When we consider consciousness, a number of questions naturally arise. Why did consciousness develop? What is consciousness good for? If consciousness developed to help us plan and act for the future, why is consciousness so difficult to control? Why is mindfulness so hard? And for that matter, if our actions are under our conscious control, why is dieting (and resisting other urges) so difficult for most of us?
Why does it appear that we are observers, peering out through our eyes at the world while sitting in the proverbial Cartesian theater? Why do we speak, in William James’s words, of a “stream of consciousness”? Can we perform complicated activities (such as driving) without being consciously aware of it?
Are animals conscious (and if so, which ones)? Are there developmental, neurologic, or psychiatric disorders that are actually disorders of consciousness?
There have, of course, been many answers to these questions over the last 2500 years. We hope to provide new answers to these and a number of related questions in this paper.
Before we attempt to answer these questions, we should clarify what we mean when we use the word consciousness. For the most part, we mean what William James (1890) meant when he used the term: our own personal experience of perceiving, thinking, emoting, and acting. Self-consciousness, that is, being conscious of our selves as thinking entities (à la Descartes), would certainly be included in what we mean, but only a small part of it. Similarly, the various so-called levels of consciousness, as measured by the Glasgow Coma Scale (Teasdale and Jennett, 1974), including conscious, confused, delirious, somnolent, obtunded, stuporous, and comatose (Posner et al, 2019), would be included in our use.
Any understanding of consciousness should also be consistent with the four basic properties that arise from studying its phenomenology: intentionality, unity, selectivity, and transience (Schacter et al, 2019). Consciousness is directed toward an object; it is about something (ie, intentionality). We experience consciousness as unified, as one experience, rather than as separate experiences of sights, sounds, smells, thoughts, feelings, and so on (ie, unity). We have the capacity to be conscious of some things and not others (ie, selectivity). Objects of consciousness are there transiently; the contents of consciousness tend to change (ie, transience).
Defining postdictive effects will be important for understanding our theory of consciousness. As counterintuitive as it sounds, in postdictive effects, a later stimulus can affect the perception of an earlier stimulus, or earlier and later stimuli can mutually affect each other (Herzog et al, 2020; Michel and Doerig, 2021; Sergent, 2018).
We wish to keep the commonsense use of the word conscious when we say, “I was so wrapped up in my thoughts that, driving without being conscious of where I was going, I found myself sitting in the empty parking lot at work, despite intending to drive to the post office.” We also want to retain the idea of having unconscious awareness, such as when we see our friend, are sure that something is different about him, but it takes us a minute to figure out that he got a new haircut. We would like to separate this idea of unconscious awareness, which may contain a conscious feeling (such as familiarity) short of a full conscious experience, from unconscious knowledge, which may influence our actions without any awareness (such as priming).
Block (2011) and others have promulgated the idea of two separate aspects of consciousness. Phenomenal consciousness is what it is like for us to have an experience. Access consciousness is when representations are made available to cognitive processing.
James (1890) famously used the phrase “stream of consciousness” as a metaphor to describe the intuitive feeling that there is not only a “now” but also “downstream” events that have occurred in the past and “upstream” events that will come to pass. Why does consciousness feel this way when we know that the brain is actually processing massive amounts of information in parallel? And does consciousness flow linearly with time, as this metaphor implies?
In fact, postdictive and other order effects have demonstrated that, at timescales <500 ms, consciousness does not flow linearly with time (Herzog et al, 2020; Michel and Doerig, 2021). Conscious awareness often occurs in the wrong order (ie, after, rather than before or with, the perception, decision, or action) (Sergent, 2018), and conscious sensations are sometimes referred backward in time (Hodinott-Hill et al, 2002; Libet et al, 1979). Consciousness is also too slow to guide many split-second decisions and actions that occur routinely when playing sports or musical instruments (Blackmore, 2017).
Experiments performed with individuals who have experienced a brain injury have demonstrated that consciousness is not necessary to perform a number of activities that we usually think require conscious awareness (Weiskrantz et al, 1974). Lastly, mindfulness is hard, which suggests that controlling our conscious thoughts is not easy to do—something that is quite odd if the purpose of consciousness is to enable us to control our thoughts and actions. We will review these problems briefly here, along with how these problems lead us to the largest problem—the purpose of consciousness.
Order Problems: Consciousness After the Perception, Decision, Action
There are many examples in which the consciousness associated with a perception, decision, or action seems to occur only after the physiological perception, decision, or action has actually occurred. This order is incompatible with the idea that perceptions, decisions, and actions are only possible when conscious awareness and thought are present.
Tolling Bells and Cocktail Parties
An oft-mentioned example is from Exner, quoted by James (1890) in The Principles of Psychology (and often attributed to James): “Impressions to which we are inattentive leave so brief an image in the memory that it is usually overlooked. When deeply absorbed, we do not hear the clock strike. But our attention may awake after the striking has ceased, and we may then count off the strokes.” The problem here is how we can become conscious of something after it has occurred. Were we conscious of it at the time that it occurred, or not?
Another common example, which almost everyone has experienced at a cocktail party, is that we hear our name spoken, our attention is suddenly focused, and we can then recall the earlier part of the sentence in which our name was mentioned (Blackmore, 2017). How is it that our consciousness can work backward to perceive the earlier part of the sentence that we were not paying attention to?
Block (2011) and others might argue that the simple answer to these two order problems is that we were phenomenally conscious of the sounds of the clock and the earlier part of the sentence, but that our access to consciousness arose only later. Although attractive, we believe that this is not the best explanation for this phenomenon.
Below we review several of the many experimentally created postdictive effects; see Sergent (2018),Herzog et al (2020), and Michel and Doerig (2021) for comprehensive reviews of these effects and their implications for theories of consciousness.
Cutaneous Rabbits. In the cutaneous rabbit illusion, we hold out one arm while looking the other way. The experimenter then taps quickly at precisely equal intervals with equal pressure five times at our wrist, thrice near our elbow, and twice near our shoulder. This produces the odd sensation as if a little rabbit were running up our arm—not three separate groups of taps (Geldard and Sherrick, 1972). There are Bayesian models that can closely replicate this illusion and thus in some sense explain it (Goldreich, 2007), but what is not explained is how the brain knows where to put the intervening taps running up the forearm from the wrist to the elbow before the elbow taps have occurred. If we think about consciousness in the ordinary sense, it simply makes no sense. Blackmore (2017, p. 40) stated the problem very well when she said:
If you stick to the natural idea that any tap (say the fourth one) must either have been conscious or unconscious (in the stream or not), then you get into a big muddle. For example, you might have to say that the third tap was consciously experienced at its correct place (i.e. on the wrist), but then later, after the sixth tap occurred, this memory was wiped out and replaced with the conscious experience of it happening half way between wrist and elbow. If you don’t like this idea, you might prefer to say that consciousness was held up for some time—waiting for all the taps to come in before deciding where to place each one. In this case, the fourth tap remained unconscious until after the sixth tap occurred, and was then referred back in time so as to be put in its correct place in the stream of consciousness.
In this illusion, consciousness simply fails to capture what is happening as it happens.
Color Phi Illusion. In his book, Consciousness Explained, Dennett (1991) discusses extensively the color phi illusion. In this illusion, a viewer watches a blue dot at the top of a frame, followed by a blank screen, and then a red dot at the bottom of the frame, all in quick succession. The viewer then reports two odd things. First, the viewer experiences a sensation of motion, as if the first spot were moving downward; second, the viewer believes that the spot changes color abruptly and in the middle of its illusory path (Kolers and von Grünau, 1976). The problem, of course, is that it makes no sense for the viewer to consciously experience the downward motion or the color change before the second dot is consciously perceived. How (and why) do these phenomena happen?
Keuninckx and Cleeremans (2021, p. 1) recently suggested that the color phi illusion may simply be related to “inherent dynamical and nonlinear sensory processing in the brain” and not related to consciousness, per se. This is an interesting idea that we discuss later in the context of our theory.
Color Fusion Effects. In this illusion, when a red disk is presented for 40 ms, the viewer sees a red disk. Yet, when a red disk is presented for 40 ms followed by a green disk presented for 40 ms in the same location, the viewer sees a single yellow disk (Pilz et al, 2013). Why does our conscious perception fuse the two colors together? How does the later presentation of the green disk interfere with the prior perception of the red disk?
Illusory and Invisible Audiovisual Rabbits. Postdictive effects can be crossmodal. In one experiment, three flashes are presented, each paired with a sound. When the sounds are repeated but the central flash is omitted, an illusory flash is perceived. Conversely, when all three flashes are present but the central sound is absent, the central flash is not perceived (Stiles et al, 2018). Why do these postdictive illusions happen?
Transcranial Magnetic Stimulation. Postdictive effects can be produced by brain stimulation. In one ingenious experiment, transcranial magnetic stimulation (TMS) pulses to the occipital cortex at various time intervals from 20 to 370 ms after stimuli were flashed altered how the stimuli were perceived. Thus, the TMS pulse itself acted as a postdictive stimulus, which had different effects depending on when the TMS pulse was applied (Scharnowski et al, 2009). How can brain stimulation several hundred ms after a stimulus alter the perception?
Motor Cortex First, Conscious Decision to Move Second
In one of the most convincing examples of this problem of consciousness occurring after decisions and actions, Dennett (1991) described an experiment that was performed by the neurosurgeon W. Grey Walter in 1963. Patients who had electrodes implanted in their motor cortex were set up with a slide projector and were told that they could advance the slides by pressing the button on a controller. However, the controller was fake—it was not connected to the projector. What advanced the slides was actually a signal from the implanted electrodes.
The patients experienced that the slide projector was anticipating their decisions. As Dennett (1991, p. 167) described, “They reported that just as they were ‘about to’ push the button, but before they had actually decided to do so, the projector would advance the slide—and they would find themselves pressing the button with the worry that it was going to advance the slide twice!” The commonsense view of consciousness tells us that the conscious decision to act precedes and causes the action itself. How do we explain this strange phenomenon in which motor actions occur before the conscious decisions to take these actions? How could the effect precede the cause?
Conscious Sensations Referred Backward in Time
In addition to order problems, there are also situations where conscious sensations are referred backward in time.
The stopped-clock illusion is one example of the multisensory illusion of chronostasis (Hodinott-Hill et al, 2002). In this illusion, the viewer makes a saccade to a clock with a second hand. As in all saccades, the perception of the visual information during the saccade is masked in order to prevent the viewer from experiencing motion blur. After the saccade is completed, the viewer focuses on the clock. The viewer’s experience is that the clock seems to be taking more than 1 second for the second hand to move. The explanation is that the sensory information of the image of the clock that the eyes receives after the fixation is projected backward in time to fill in the time period when the viewer was making the saccade (Thilo and Walsh, 2002). But, how can a sensation be projected backward in time?
Stimulation of the Hand Versus the Somatosensory Cortex
Following up on a prior experiment in which they produced conscious sensation by stimulating the cortex directly (Libet et al, 1964), Libet et al (1979) produced the conscious experience of a tingle in a participant’s hand by either stimulating the back of the hand or stimulating the contralateral somatosensory cortex of the brain directly. If we assume that the conscious experience is related to the somatosensory cortex receiving the stimulation—because the impulses initiated in the hand need to travel from the hand through the wrist, forearm (radius or ulnar nerve), arm (brachial nerve), shoulder (brachial plexus), neck (spinal nerves and cord), and head (including the brain stem, internal capsule, and corona radiata)—we would expect that stimulation of the cortex would be noticed more quickly than stimulation of the hand. Two surprising results were found. First, in each case, it took a long time, ∼500 ms, from stimulus onset until conscious experience. Second, as Libet and colleagues (1979, p. 222) wrote: “After delayed neuronal adequacy is achieved, there is a subjective referral of the sensory experience backwards in time so as to coincide with this initial ‘time-marker.’”
What does it mean for a conscious experience to be referred “backwards in time”? Many scientists and philosophers have provided explanations for these results and the 1979 conclusions by Libet and colleagues (Churchland, 1981; Dennett, 1991). We have no evidence that any of them are wrong; we simply think that our theory provides a more parsimonious explanation of these observations. As we will see, backward referral in time does not pose a challenge for our theory.
Timing Problems: Consciousness Is too Slow
The experiment by Libet and colleagues (1979) raises another problem of consciousness that Blackmore (2017) stated explicitly: Consciousness is too slow. Recall that Libet and colleagues (1979) discovered that it took ∼500 ms from stimulus onset until the conscious experience occurred. Blackmore (2017) reminded us just how very long that amount of time actually is from a neurophysiological perspective, where impulses travel at speeds up to 100 m/seconds. If it takes 500 ms (long enough for an impulse to travel up to 50 m) for conscious experience to occur, then consciousness is too slow to be playing an active, controlling role in many activities, including playing sports and making music.
It is estimated that professional baseball players need to decide whether to swing at a pitch within 125 ms after it has left the pitcher’s hand, and the ball crosses the plate within 300–400 ms after it has left the pitcher’s hand (Science Non Fiction, 2016). Ordinary reaction time measured by clicking a mouse in response to an auditory, tactile, or combined auditory/tactile stimulation produced speeds in one study ranging from ∼210 to 320 ms for most people, but was as fast as 100–210 ms in trained musicians (Landry and Champoux, 2017). How can we be consciously in charge of our actions if our actions are occurring much faster than our conscious thoughts?
Individuals with neurologic disorders may help us to understand consciousness. Some individuals with brain lesions are unable to consciously perform a task but, when asked to do the task unconsciously—guess or just perform the task without thinking about it—they are able to do it.
Visual Apperceptive Agnosia
In a study by Ganel and Goodale (2019), a patient with visual apperceptive agnosia was unable to consciously perceive and report the size or shape of objects, yet she was able to correctly scale her grip to pick them up. She was also able to accurately perform actions, such as putting cards through slots of different angles, despite not being able to consciously perceive or report the angles. How is this performance without conscious awareness possible? Moreover, this dissociation between inaccurate visual perception and accurate grip scaling can be produced in individuals with normal vision as well (Aglioti et al, 1995; Chen et al, 2015); again, how is this possible if our actions are consciously controlled?
Blindsight causes problems for consciousness similar to those observed in patients with visual apperceptive agnosia (Kentridge et al, 2008; Poppel et al, 1973; Weiskrantz et al, 1974). Individuals with blindsight are cortically blind in at least one hemifield; cannot consciously see objects in that hemifield; and yet perform above chance when they are asked to guess, point, or otherwise act based on a visual stimulus to their blind field. Explanations of the physiological phenomenon include preserved cortical islands; dissociation between a dorsal, action-related unconscious stream and a ventral, perception-related conscious stream (Brogaard, 2011); and visual information reaching subcortical structures such as the superior colliculi and lateral geniculate nuclei. These physiological explanations, however, still fail to explain the phenomena regardless of its biological underpinnings. How is it possible to accurately point at objects and perform similar tasks when such objects cannot be consciously visualized?
Mindfulness is a problem because it is hard. Anyone who has tried to practice mindfulness knows that it is difficult. But if, as a commonsense view of consciousness would have it, our consciousness (ie, access consciousness) evolved to allow us to perform high-level abstract reasoning using language, logic, visuospatial abilities, or other cognitive capacities in order to enable us to carry out intentional actions, then controlling our thoughts should be easy. So why is mindfulness so hard? Why is it so difficult for us to control our conscious thoughts?
Lack of Apparent Causal Role for Consciousness
The idea that it is difficult to control our thoughts, and that individuals with brain lesions and no relevant conscious perception can still perform tasks that we intuitively feel must require consciousness, leads to the interesting—and for most people, uncomfortable—thought that perhaps consciousness is epiphenomenal. Many researchers (eg, Chalmers, 2010) have made this point, arguing that there could be philosophical zombies who act as if they are conscious but actually are not. Some biological traits will survive through natural selection only incidentally, because they arose in association with an adaptive trait (Gould and Lewontin, 1979). Is consciousness like that, merely epiphenomenal? Perhaps consciousness appears to be epiphenomenal only because we are looking in the wrong place to find its causal role. Could consciousness be epiphenomenal with respect to perceptions, decisions, and actions, but play an important causal role in some other function?
How Does Consciousness Contribute to Evolutionary Success?
To review, consciousness often occurs after the perception, decision, and action have taken place, in part because consciousness is too slow to participate in many real-time events. Consciousness fools us in many different ways, creating visual and tactile illusions. We know from individuals with various brain lesions that a variety of judgments and actions can occur without conscious perception, suggesting that although consciousness is commonly present, it is not necessary to carry out at least some types of tasks. Lastly, consciousness is difficult to control—something that seems very odd if it evolved to allow us to carry out the complex reasoning that is necessary for intentional action.
All of these problems lead us to one of the most important questions regarding consciousness: What does consciousness do? How does consciousness contribute to the evolutionary success of human beings?
CONSCIOUSNESS AS A MEMORY SYSTEM
The problems discussed thus far demonstrate just how many difficulties there are if we simply take at face value the idea that the role of consciousness is to enable our perceptions, decisions, and actions to occur. In this setting, we are ready to explain our theory that consciousness is, at its core, a memory system.
Consciousness Is Part of the Episodic Memory System
Tulving (1985) described episodic memory as the set of processes that allow us to mentally time-travel and to re-experience a past moment in time. To be able to do that, we must first take information that comes in through our sensory stores and our working memory and then create a mental representation of a moment in time; this is the process of encoding. If we want that representation to be accessible later, we must store it in some durable form; this is the process of consolidation. And, when we want to later reflect on that moment in time, we must engage retrieval processes to do so.
We argue that one function of consciousness—and more importantly, what it initially developed to do—is to allow for each of these phases of episodic memory. Consciousness binds elements of an experience together, allowing for the creation of a memory trace that can include multisensory details. Over time, consciousness provides a medium in which these memory traces can be replayed—a mechanism that is key to their successful storage.
This idea has been hinted at or suggested in some form for decades (Dafni-Merom and Arzy, 2020). In 1985, Tulving (p. 2) wrote, “Remembering is a conscious experience. To remember an event means to be consciously aware now of something that happened on an earlier occasion.” The author further explained that different memory systems are characterized by different kinds of consciousness: procedural memory by anoetic (nonknowing) consciousness, semantic memory by noetic (knowing) consciousness, and episodic memory by autonoetic (self-knowing) consciousness.
In 1995, Moscovitch (p. 1341) wrote, “Consciousness is an inherent property of the memory trace, being bound to it along with other aspects of the experienced event by the hippocampus and structures related to it. … With respect to remembering, and perhaps with respect to no other function, consciousness is also an inherent property of the very object of our apprehension.” However, as these examples show, the relationship between episodic memory and consciousness is normally explained as the other way around, that episodic memories bind conscious experiences together, not—as we are suggesting—that the conscious experience is the process of remembering.
Cleeremans (2011) also suggested in his radical plasticity thesis that learning and memory are necessary for consciousness to develop. In his account, the brain “learns” to be conscious by continuously attempting to predict the consequences of its actions on both its self and the outside world. This activity produces meta-representations that, when combined with the emotional value associated with them, produce conscious experience. Thus, although learning, memory, and plasticity are necessary for consciousness in Cleeremans’s thesis, his views are clearly quite different from our theory.
Episodic memory is now understood to have value not just for its ability to represent the past, but also for its utility in allowing the use of past experiences to increase understanding of the present moment and to make predictions about future occurrences (Schacter et al, 2007; Suddendorf and Corballis, 2007). At retrieval, these traces of episodic events can be flexibly and creatively combined and rearranged in consciousness to allow anticipation, future planning, and intentional action. If we view consciousness as having developed as a critical part of the episodic memory system, we believe that all of the questions that we asked can be answered, and that all of the problems that we raised can be solved—or they simply cease to look like problems.
To elaborate, our theory of consciousness rejects the idea that consciousness initially evolved in order to allow us to make sense of the world and act accordingly, and then, at some later point, episodic memory developed to store such conscious representations. Our theory is that consciousness developed with the evolution of episodic memory simply—and powerfully—to enable the phenomena of remembering. We view the fact that our ability to imagine things in consciousness is constrained by and related to our episodic memory (James, 1890; Moulton and Kosslyn, 2009) as another piece of evidence supporting the idea that consciousness evolved as part of episodic memory.
Today, consciousness certainly participates in functions that we do not generally associate with episodic memory, such as problem-solving, abstract reasoning, and language. We argue that such functions developed later in evolutionary history, after consciousness was already functioning to furnish the content of episodic memory representations.
Consciousness, Sensory Memory, Working Memory, Episodic Memory, and Semantic Memory Are Part of the Same System
At this point, we would like to clarify the distinction between consciousness and working memory—the ability to keep information in mind and manipulate it. Our theory is entirely consistent with the ideas put forth by Baddeley, Hitch, and others regarding the phonological loop, visuospatial sketchpad, central executive, and episodic buffer (Repovs and Baddeley, 2006). Moreover, we would argue—along with many others—that although we can split sensory, working, episodic, and semantic memory into separate systems, in the healthy brain, these systems function seamlessly together as a single system (Renoult et al, 2019; Repovs and Baddeley, 2006).
Information of which we are consciously aware either from our senses, via sensory memory, or from long-term memory stores comes into working memory. For example, let’s say that we are walking in our neighborhood when we hear a bark. This auditory information is present in our sensory memory for mere seconds, but it is long enough for us to transfer it to our working memory. Once in our working memory, this auditory information acts as a cue that triggers the retrieval of an episodic memory from last week—when the dog that makes that particular bark chased us to the edge of its property! Now that the sensory information of the bark and last week’s experience of being chased are together in our working memory, it is not difficult to imagine the future: The dog is likely to chase us again. Without waiting for the dog to chase us, we quickly cross the street.
So we have sensory memory, working memory, and episodic memory helping us sense, remember, imagine, and act. But, where does consciousness come in? We argue that we hear the bark when we are consciously aware of it. We consciously remember last week’s chase when we retrieve the elements of the prior experience and build a conscious representation of that prior experience in working memory. Without effort, we consciously imagine the dog chasing us again. And—either consciously with deliberation or automatically and unconsciously without deliberation—we walk across the street. In this example, we view consciousness as being necessary for the sensory memory perceptions, retrieval of episodic memories, and imagination of the future that may have led to our action of crossing the street. We view consciousness as an integral part of sensory memory, working memory, and episodic memory, and in that way, we view all of these elements as part of the same memory system.
Now, we will note that we do not really need consciousness for the action we performed of crossing the street. In fact, we might correctly note that we do not need episodic memory either. We just need operant conditioning to hear the bark and walk across the street. This is one reason we chose this particular example. There are two points we would like to make here.
First, this is yet another example where, at least at a superficial level, consciousness does not seem to be necessary. So why is consciousness present? We now have an explanation that we did not have when we first discussed this problem earlier: Consciousness is necessary to remember this event. Could consciousness be epiphenomenal in relation to the dog barking and our walking across the street? Yes. But, it could not be epiphenomenal in relation to our forming and retrieving an episodic memory; thus, we argue that consciousness is essential.
Second, we believe that our conscious episodic memory record of the dog barking is critically important in many situations. One is when the stimuli and setting are not similar enough to trigger the conditioning response. Another is when the conditioning response is triggered but the situation requires us to choose actions different from what we are conditioned to do. In both of these situations, the ability to consciously retrieve different memories and to imagine different scenarios is critical for making decisions that could have evolutionary implications, as pointed out by both Schacter et al (2007) and Suddendorf and Corballis (2007). In fact, individuals with impaired episodic memory show impairment in their ability to learn from their prior errors despite showing near-normal implicit memory (Baddeley and Wilson, 1994), again demonstrating that conscious episodic memory is critical for learning that leads to flexible future performance.
Now, it would certainly be useful to be able to generalize our experience with this particular dog to other dogs, so that, if we are taking a shortcut through a different neighbor’s yard and their dog begins to bark, we will know that we should move swiftly to get off their property even before the dog starts chasing us. This ability to form general facts from specific events stored in episodic memory is, of course, one way to define semantic memory. Here, we would simply like to make the point (that many others have as well [eg, Janssen et al, 2022; Renoult et al, 2019]) that although we can view semantic memory as a separate memory system, we can certainly think of it as being part of the same system as episodic memory.
Our View Differs From Baars’s Global Workspace Theory
Next, we would like to comment on the global workspace theory of consciousness that was proposed by Baars (2005) and expanded on by others (eg, Gaillard et al, 2009). To explain this theory, Baars (2005, p. 46) used a metaphor of a theater in which consciousness “resembles a bright spot on the stage of immediate memory, directed there by a spotlight of attention under executive guidance.” Although our theory of consciousness is mostly consistent with global workspace theory, the major difference is that our theory adds the original purpose of consciousness: to allow us to store prior experiences in episodic memory.
Baars (2005) outlined six theoretical claims; our comments on these claims are as follows. In claim 1, Baars suggested that conscious perception enables access to widespread brain sources, which we agree with. However, he continued, stating that compared with conscious perception, unconscious sensory processing is more limited, which we do not believe has been proven or disproven; our suspicion is that it is not more limited than conscious perception. Claim 2 is that conscious processes enable working memory and that there is no evidence for unconscious working memory. We agree that part of the definition of working memory is that it is conscious. Claim 3 begins by stating that consciousness enables episodic and other forms of explicit learning, which we certainly agree with. However, it then continues to state that conscious events also enable implicit and skill learning, which we believe is only partly true. It is true that many instances of skill learning and other forms of implicit memory are enabled or facilitated by conscious events, such as taking a tennis lesson or practicing the viola. There are, however, a number of instances of implicit learning, such as priming, in which the learning occurs without consciousness.
Claim 4 of Baars’s (2005) claims starts by stating that conscious perceptual feedback facilitates voluntary control over motor functions. We believe that this is sometimes true—such as when we are working on improving our backhand or our vibrato—but note that, in our view, the vast majority of motor actions are unconscious, and thus the perceptual feedback will also be unconscious. Claim 4 continues by speculating that conscious perceptual feedback enables voluntary control over neuronal populations and perhaps single neurons. On the one hand, if we are consciously changing our behavior, there will necessarily be changes in the brain; this seems obvious unless one is a dualist. On the other hand, whether it is single neurons or assemblies of neurons that are changed is an empirical matter that future experiments can resolve.
Claim 5 is that attention can be directed by conscious thought but can also be captured by relevant stimuli (such as a name, the need to use the toilet, or a fire-alarm signal), which then becomes the contents of consciousness. We certainly agree with this claim. Claim 6 has phenomenological and brain components. The phenomenological claim is that consciousness enables access to the observing self through executive interpreters. This claim makes intuitive sense, although it seems to evoke a homunculus sitting in the Cartesian theater. Claim 6 goes on to state that this process involves the parietal and prefrontal cortex; we agree that these brain regions are certainly important, and we will discuss them in more detail later.
Lastly, we reiterate that, in addition to these differences between global workspace theory and our theory, the major distinction is that our theory adds that the original purpose of consciousness was (and a major purpose still is) to facilitate the encoding, storage, retrieval, and flexible recombination of prior events using episodic memory and its related memory systems.
Attention Is Necessary but not Sufficient for Conscious Awareness
Insofar as what we attend to determines which contents are engaged in working memory and thus have the potential for being stored as episodic memory, we believe that what we are conscious of (ie, the object of consciousness) depends on what we are attending to (ie, the object of attention). Therefore, we believe that attention is necessary for conscious awareness.
Attention is not, however, sufficient for consciousness, as has been demonstrated through many experiments that have manipulated both voluntary (endogenous) and external, reflexive (exogenous) attention and produced a change in performance and/or reaction time without conscious perception (Hsieh et al, 2011; Kentridge et al, 2008; Zhang et al, 2012). For overviews, see Breitmeyer (2015) and Breitmeyer et al (2015).
There are also examples where attention to one stimulus can cause other stimuli to disappear or be absent from conscious perception. The Troxler fading effect shows that attention that is directed to one area of a visual scene can cause effects of either fading or intensifying to other areas of the scene, which may be due to microsaccades (Alexander et al, 2021). In motion-induced blindness, relevant targets may disappear when they are viewed against a moving background (Thomas et al, 2017). In change blindness (Andermane et al, 2019), inattentional blindness (Hutchinson et al, 2021, and the attention blink (Sy et al, 2021), even highly incongruent stimuli—such as someone in a gorilla costume walking through a basketball court (Simons, 2010)—can be invisible to conscious perception.
We believe that all of these phenomena are consistent (or, at least, not inconsistent) with our theory, which we will refer to as our memory theory of consciousness. To reiterate, attention is necessary, but not sufficient, for stimuli to enter working memory. Therefore, stimuli that are unattended to will neither be consciously perceived nor remembered using working or episodic memory. If these unconsciously perceived stimuli are learned, it occurs with unconscious memory processes, such as priming.
Our Theory Is Consistent With the System 1 (Unconscious) and System 2 (Conscious) Distinction
Our memory theory of consciousness is fully consistent with the distinction that Kahneman and Tversky (Kahneman, 2011; see also Carruthers, 2015) made between the slow, effortful, logical, calculating, conscious System 2 and the fast, automatic, stereotypic, unconscious System 1. Our theory would simply add that conscious System 2 was made possible by the original purpose of consciousness—to be the contents of episodic memory.
From the Contents of Memory to Problem-solving and Abstract Reasoning
How did consciousness move from being solely the contents of episodic memory to being involved with problem-solving, abstract reasoning, and the other abilities made possible by System 2? We speculate that consciousness evolved and became involved in these other abilities due to consciousness being a key element of the episodic memory system’s function of flexibly and creatively combining episodic memories to imagine the future.
We envision an early stage of this type of combination that would simply allow us to predict the future. For example, episodic memories of finding a delicious berry near a specific cave each autumn might combine with another episodic memory of being chased by a bear near that cave. The outcome is that we are able to predict that if we go to pick the berries, we might end up being chased by a bear.
In addition to merely envisioning how the future might unfold, at some later point, this conscious memory creative recombination process envisioned two or more possible future outcomes. In one future, when we go to pick the berries, we are chased by the bear, whereas in another future, we are not chased.
Once consciousness is able to compare two possible futures, problem-solving comes in when we think about what we can do to help bring about the future that we want and to avoid the future that we do not. For example, memories of ways to determine whether an animal is in its lair or not may come to mind. Other memories may remind us that a bear can only chase one person at a time. Comparing these types of memory retrievals—that perhaps we may now refer to as thoughts—can allow a plan to develop. Problem-solving in consciousness/working memory via episodic memory retrieval is now taking place.
Once problem-solving is occurring, it is a small step to conscious abstract reasoning. As multiple events stored in episodic memory generalize into semantic memory, abstractions automatically occur. Episodic memories of individual dogs, bears, and rabbits allow for the general semantic memory categories of dog, bear, and rabbit to be formed. A more abstract semantic memory category of animal can then emerge and be contrasted with the abstract semantic memory category of plant, and so on.
Much has been written about the development of language that will not be repeated here (eg, Pinker, 1994). Succinctly, we believe that language developed from a conjunction of consciousness and semantic memory.
One of the things that makes language an interesting case, however, is that although we can certainly speak with full conscious awareness and deliberation, it is our observation that we can also speak unconsciously, without thinking about it. We will return to this important concept in a later section. For now, we simply want to introduce the idea that just because a function developed with consciousness does not mean that it must be present only with consciousness.
Conscious Perception as a Memory
At this point in our paper, one might be willing to accept our theory that consciousness evolved as part of the episodic memory system but say, “So what? How does this explanation help us understand consciousness (or, for that matter, episodic memory)?”
If we believe that consciousness evolved as part of the episodic memory system, as a critical part of that system that allows us to store prior experiences in memory and retrieve them so that that the memories of these experiences can be flexibly and creatively combined to allow future planning and intentional action, then there is no reason that consciousness needs to operate in real time. If consciousness is a system for memory encoding and retrieval—and not direct action—there is no reason that it cannot function properly with a small delay. We would argue, in fact, that we do not consciously perceive events directly in real time. We perceive the world as a memory. In other words, technically, we are not consciously perceiving anything directly; we are actually experiencing a memory of a perception.
We suggest that we experience the world by remembering sensory memories. Moreover, most of the time, we are not experiencing these bottom–up sensory memory processes by themselves. We experience sensory memory processes influenced by top–down episodic and/or semantic memory processes, such that the percept that is consciously perceived is a mashup between the bottom–up sensory memory processes and the top–down episodic and sematic memory processes.
Postdictive Effects Explained
Our memory theory of consciousness can now explain postdictive effects. For example, Sergent et al (2013) showed that not only does cuing a stimulus before its presentation improve conscious perception of the stimulus, but cuing after the stimulus presentation can as well. Sergent and colleagues concluded that (p. 154):
the initial sensory processing associated with a stimulus can occur preconsciously, because its conscious or nonconscious fate can change drastically beyond this phase. Conscious perception would thus relate to secondary amplification of preconscious information held in sensory areas. … this secondary amplification does not have to be a direct consequence of the initial processing of the stimulus itself, but can be triggered by a subsequent and independent event.
Our memory theory of consciousness is both consistent with this explanation and can allow us to understand it better. Attention is drawn to the unconscious perception by the poststimulus cue, and we then experience the conscious perception in the same way that we experience all conscious perceptions—by remembering it.
The Conscious Memory of Unconscious Decisions and Actions
If we are willing to consider that we—at least as conscious selves—do not perceive the world directly but rather remember it, then we are ready to explain conscious decisions and actions. Our theory is simply that the brain processes that decide and act are unconscious. Our conscious decisions and conscious actions are actually memories of those unconscious decisions and actions. We believe that this explanation—that decisions and actions are fundamentally occurring through unconscious brain processes—is consistent with an evolutionary perspective that would argue that there is no single conscious decision-making system in the brain. Instead, there are various processes that are unconsciously engaged to make specific decisions, such as when to eat, sleep, avoid, approach, grasp, release, and so on (Cisek, 2019).
We do think, however, that unconscious brain processes will sometimes engage the conscious memory system to facilitate optimal decision-making and performance in certain situations. To explain these notions further, we will provide an example using Kahneman and Tversky’s System 1 and System 2 processing (Kahneman, 2011; see also Carruthers, 2015).
System 1 Decisions and Actions
Let’s first consider System 1 decisions. These are the fast, automatic, unconscious decisions that require little or no thought or effort. Let’s say that we are working, perhaps engrossed in thought writing a paper on our computer. We suddenly decide, “I’d like a glass of water.” Before we finish typing the paragraph we are working on, and still thinking about the sentence that comes next, we rise, walk into the kitchen, open the cabinet door, reach inside, grasp a glass, pull it out, close the cabinet door, hold the glass under the faucet, turn the cold water on, watch the glass fill, turn the water off, raise the glass to our lips, tilt the end upward, take a sip, and, while holding the glass steady so as not to spill, walk back to our computer, set the glass down, sit down, and get back to work.
We would argue that our decision to get a glass of water could well have been unconsciously initiated and acted out, as could every part of the sequence. In fact, we believe that almost all of the time when we walk or grasp an object, completely unconscious brain processes carry out these actions (consistent with Aglioti et al, 1995; Chen et al, 2015; and Cisek, 2019). Insofar as we are consciously aware of what we are doing, we suggest that this conscious awareness is a memory of this decision and action. On this account, consciousness is epiphenomenal with respect to the decisions and actions, but not epiphenomenal in general because it plays an important role via the episodic memory system.
System 2 Decisions and Actions
Now let’s consider some analogous System 2 decisions and actions. Perhaps, instead of working on our computer, we are participating in a dangerous Hunger Games-like activity. We are thirsty, and we can see the cool spring up ahead. But to reach it, we need to either make our way across a bubbling lava field with floating rocks that we have to step on or run across a grassy meadow filled with poisonous snakes and spiders. We carefully consider our options. In the end, we decide that we will have a better chance jumping from floating rock to floating rock across the lava. We carefully step on one rock, get our balance, and wait for another to float nearby. We time our jump perfectly and land in a crouch, distributing our weight. We continue this way until we reach the spring. We can see the glass we need, but to get it, we need to carefully reach our hand through razor-sharp barbed wire. We contort and angle our hand and fingers to reach the glass and delicately pull it back through. We fill our glass from the stream and drink down the precious liquid.
Here, we have a series of decisions and actions that need to be thoughtfully and carefully carried out. Instead of acting automatically (perhaps while thinking about something else), we have to consciously consider and fully attend to each of these decisions and actions. However, we believe that the actual decisions and actions themselves are made and carried out by our unconscious self, and that we experience the conscious decision being made or action taking place only after the fact.
Unconscious Perceivers, Decision Makers, and Actors
Another way of saying this is that it is the System 1 unconscious parts of our brain that actually perceive, make decisions, and act (consistent with Cisek, 2019), and the System 2 conscious parts that provide an additional layer of information that our unconscious brain can use (or not) to make decisions and act accordingly. System 2 uses consciousness, and thus all of the explicit memory systems, to review what we know about lava flows, snakes, and spiders (from semantic memory); how well we did the last time we had to jump from rock to rock (from episodic memory); and how, by counting in our head and watching the floating rock (using working memory), we might be able to time a jump perfectly.
The Conscious Memory System
Thus far, we have sometimes been using the term episodic memory in its standard definition (ie, memory for prior events) and sometimes as shorthand for all of the explicit memory systems: working memory, episodic memory, and semantic memory. Because we believe that all of these explicit memory systems are truly part of one system—the explicit or conscious memory system—going forward, we will use the terms episodic memory and episodic memory system just in their narrow senses and the terms conscious memory and conscious memory system in this broader sense, referring to all explicit memory systems.
Challenges for Our Theory
If we are correct that consciousness and explicit forms of memory are all part of the same system, then there should not be examples of explicit memory that are unconscious. We could immediately argue that there cannot be examples of unconscious explicit memory, as that would be the same as saying that there are examples of unconscious conscious memory. Nonetheless, it has been shown through carefully conducted experiments that unconscious processes similar to episodic memory, using similar anatomical networks, have enabled participants to perform inferences that would usually require conscious awareness to perform (eg, Schneider et al, 2021). In addition, other carefully conducted experiments have shown unconscious (ie, guessing) above-chance performance on delayed-response working memory tasks that would usually require keeping information consciously in mind (eg, Trübutschek et al, 2017). Do these experimental results mean that our memory theory of consciousness must be false? Although we readily admit that such experimental results are problematic for our theory and need to be explained, we would argue that they do not negate it, for the reasons presented in the next two sections.
Unconscious Episodic Memory?
In the Schneider et al (2021) study, participants were exposed to either weak masking of stimuli that allowed for conscious processing or strong masking of stimuli that necessitated unconscious, subliminal processing. Conscious processing led to an improvement in reported accuracy as well as a reduction in reaction time. Unconscious processing led to a reduction in reaction time but did not improve accuracy, which was at chance. Further, this reduction in reaction time for unconscious processing was only observed in “intuitive” decision makers who habitually responded according to their instincts (using System 1), and not in “deliberative” decision makers who preferred relying on consciously accessible knowledge (using System 2). The fMRI experiments that showed the neuroanatomical correlates of this unconscious processing were conducted exclusively with intuitive decision makers.
Our first comment is that it is possible that even the strongly masked stimuli were minimally or partially conscious because it is difficult to exclude this possibility in experiments of this type (eg, Holender, 1986; Timmermans and Cleeremans, 2015). Thus, one explanation of these results is that the strongly masked stimuli engaged the episodic memory system because the stimuli are minimally or partially conscious for some individuals.
Our second comment is that if we agree that the strongly masked stimuli are processed unconsciously, it is possible that they still do engage the episodic memory system, but only partially, and not strongly enough for a full, true, conscious episodic memory to be formed. Support for this view comes from the fact that these strongly masked stimuli produced a change in reaction time, but not a change in accuracy. Thus, we would argue that although the strongly masked stimuli unconsciously activated the episodic memory network and produced a change in reaction time, there was no change in accuracy, no true episodic memory was formed, and therefore, there was no requirement for consciousness.
Unconscious Working Memory?
In the Trübutschek et al (2017) study, participants identified the location of visual stimuli after a delay. Stimuli were rated by the participants on a 1 (unseen) to 4 (clearly seen) scale. The participants were instructed to guess at the location even if they were unable to see the stimulus. Behaviorally, the participants performed greater than chance on both the seen and the unseen trials. Magnetoencephalographic data were also obtained in order to determine if the same or different neural mechanisms were used by the participants when identifying the seen (correct) trials versus the unseen (correct) trials, with the hypothesis that if different neural mechanisms were engaged, accuracy on the unseen (correct) trials was not simply due to the participants misclassifying trials as unseen that were, in actuality, glimpsed. The magnetoencephalographic data clearly showed two different neural mechanisms.
Trübutschek and colleagues (2017) discovered that conscious and nonconscious working memory used different brain mechanisms, and that nonconscious working memory used an “activity-silent mechanism” based on slowly decaying calcium-mediated synaptic weights. The authors then postulated that perhaps this activity-silent mechanism underlies both conscious and nonconscious working memory, which they supported with modeling (but not empirical data).
Our comment on this study is simply that there are many examples of unconscious processing that leads to a future change in behavior or performance; priming and procedural memory being two. Thus, although we do not dispute the findings that there may be an activity-silent mechanism that supports unconscious processing of a spatial delayed response task, we would suggest that calling this processing nonconscious working memory may not be using the best nomenclature.
ANSWERS AND SOLUTIONS
Let’s now review the many previously inexplicable findings discussed earlier and consider how our memory theory of consciousness can provide explanations for each, along with some additional inferences.
Order Problems: Consciousness After the Perception, Decision, Action
If the contents of our consciousness (ie, what we are consciously aware of) are a memory of the perception, decision, and action, then there is no difficulty with consciousness occurring after the perception, decision, and action.
Tolling Bells and Cocktail Parties
Thus, there is no difficulty with our being able to count off the strokes of the clock even though we did not pay attention to them until the last chime; our conscious awareness is always a memory of the chimes. Similarly, it is not surprising that when we hear our name at a cocktail party, that we can hear the earlier part of the sentence; we are remembering it, just as we do with all of our other experiences. This feature of our awareness just becomes more apparent to us when we reflect on certain types of situations.
Motor Cortex First, Conscious Decision to Move Second
The experiment conducted by W. Grey Walter now makes sense as well. In normal individuals (without electrodes implanted in their brain) who are controlling a slide show with a carousel and a push-button controller, their experience is that they make a conscious decision to advance the slide, consciously use their thumb to push the button, and then, the slide advances. However, what actually occurs is that the individuals make an unconscious decision to advance the slide—then consciously remember that unconscious decision, unconsciously use their thumb to push the button—then consciously remember that unconscious action, and then the slide advances. The conscious memory for the decision and action are “time-stamped” by the brain to occur not only in the proper order, but also at the proper time such that it appears that this conscious decision and conscious action coincided with the unconscious decision and unconscious action even though the conscious memory of these events was actually experienced after the events themselves. This is not strange or mysterious: It is the nature of memory that remembered events are referred to a previous time.
In the case of the patients with electrodes implanted in their motor cortex, the patients unconsciously decide to advance the slide—then consciously remember that decision. They unconsciously decide to use their thumb to push the button, which creates the motor cortex readiness potential that triggers the slide to advance, the slide begins to advance, they consciously remember the slide beginning to advance, then consciously remember pushing the button, generating this feeling of the slide projector anticipating their decisions.
In an experiment related to motor movement, Libet (1985) compared the time in which participants consciously decided to move their wrist (determined by participants noting the time they made their decision on a special clock) with the measured readiness potential of this voluntary action at the scalp. The readiness potential has been interpreted as representing the final stages of planning preparation for movement. What was startling was that Libet (1985) found that the readiness potential preceded the voluntary decision by ∼350 ms. The author concluded that the initiation of a spontaneous voluntary act “begins unconsciously.”
We did not introduce this important experiment before now because it is quite controversial for at least two reasons. The first is that not everyone who has tried to replicate the experiment has been able to do so, although some have (eg, Vinding et al, 2014), which should set that issue to rest. The second is that Schurger et al (2012) conducted a terrific set of related experiments and analyses that they suggest explains the readiness potential not as representing the final stages of motor preparation, but rather as representing spontaneous subthreshold fluctuations in neuronal activity.
We do not disagree with the work done by Schurger and colleagues (2012), as their explanation may be the correct one for this phenomenon. We would simply like to point out that our memory theory of consciousness makes Libet’s (1985) initial interpretation comprehensible: Decisions and actions are initiated unconsciously and then we experience the conscious memory of those decisions and actions. If we believe Libet’s (1985) interpretation, and these motor results are generalizable, our conscious memory for our decisions and actions may occur ∼350 ms after the decisions and actions are unconsciously initiated.
The Incredible Slowness of Consciousness
We can now understand why it does not matter that consciousness is too slow for the real-time decisions and actions of athletes, musicians, and others who need to react quickly. All decisions and actions are occurring unconsciously. Our conscious memory of these decisions and actions occurs later.
Conscious Sensations Referred Backward in Time
Stimulation of the Hand Versus the Somatosensory Cortex
Our memory theory of consciousness also helps us to understand some of Libet and colleagues’ (1979) other experimental results and how these results are informative regarding the timing of consciousness. Recall that the authors found that it took ∼500 ms of cortical stimulation before conscious experience of a tingle occurred. If the stimulation was <500 ms, the participants did not report feeling anything. Libet and colleagues (1979, p. 222) referred to this extended time as “the neuronal adequacy for consciousness.” They explained that we are not aware of this delay because events are referred backward in time after neuronal adequacy has been achieved. The idea is that when we feel a touch on our arm, the impulses travel through our peripheral nerves, spinal cord, and brain until they reach the somatosensory cortex. If the stimulus activates our cortex for at least 500 ms, we consciously feel the touch, and it is referred backward in time such that we do not notice any delay in the conscious perception (Blackmore, 2017; Dennett, 1991; Libet et al, 1979).
The first point to make here is that our new understanding of consciousness—that we do not perceive events directly but only remember them—makes it easy to understand this finding of Libet and colleagues (1979): Once the 500-ms threshold for conscious sensation is crossed, we can simply remember the sensory memory of the touch. It is not even surprising that it is referred backward in time—again, it is fundamental to memory that it allows us to experience events that occurred earlier.
The second point that Libet and colleagues’ (1979) experimental results could suggest is that 500 ms is the amount of time that our conscious perceptions are delayed. In other words, if this result from these authors is generalizable to other conscious sensory experiences, then our conscious perception of the world may be delayed by half a second—although referred backward in time so that we do not notice the delay.
The problem of the stopped clock illusion is that the final fixation of the clock is projected backward in time to fill in the time period when we were making the saccade. This is no longer a problem because our visual perceptions are not directly experienced; they are consciously remembered. In other words, we are consciously perceiving a memory that is delayed ∼500 ms. Thus, a top–down process in our visual system projects the final fixation backward in time in order to fill in what would have been the missing perceptual experience while the saccade was taking place. Problem solved.
Because conscious perception is a memory, likely delayed by ∼500 ms and referred backward in time, we now have an easy and comprehensible explanation of postdictive effects.
In the case of the cutaneous rabbit, there are five taps at the wrist, three near the elbow, and two at the shoulder, yet we consciously experience intervening taps as well. Again, because sensations are consciously remembered (500 ms later) rather than consciously experienced, it is not a problem for some top–down process to interpose, backward in time, intervening taps between our wrist and elbow and between our elbow and shoulder such that we consciously perceive a little rabbit running up our arm. An analogous explanation can explain the illusory and invisible audiovisual rabbits.
Color Fusion Effects
Color fusion effects can be explained in a similar fashion. When a red disk is presented for 40 ms by itself, there is the obligatory delay, and then we consciously perceive the red disk by remembering it. When the red disk is followed by the green disk in the same location, the colors are fused together in our sensory memory (because they are occurring in the same time window) and, after the 500-ms delay, we perceive the fused yellow image by remembering it.
It should no longer be surprising that a TMS pulse between 20 and 370 ms can induce postdictive effects. Because conscious perception is delayed, if there is disruption to the visual stream during the delay period, it can alter conscious perception of the prior stimulus just like a physical stimulus can.
Color Phi Illusion
Similarly, we can now explain why it is that when we watch the color phi illusion and see a blue dot at the top of a frame followed by a blank screen, and then a red dot at the bottom of the frame, we consciously experience the blue dot traveling down and changing color before we see the red dot. Again, a top–down process has interposed, backward in time, the intervening dots and the color change into our conscious perception—easy to do because that conscious perception is actually a memory.
What about the suggestion by Keuninckx and Cleeremans (2021, p. 1) that the color phi illusion may simply be related to “inherent dynamical and nonlinear sensory processing in the brain” and not related to consciousness, per se? We would argue that our work helps to elucidate one part of this phenomenon (how events are referred backward in time), while theirs helps to elucidate another (why there is a sense of motion and color switch before the second stimulus is seen).
Creating Uniformity: Remembering the Gist
Our memory theory of consciousness may also help to explain why we do not notice that our peripheral vision is grayscale or that our vision is filled with the blobs and stripes related to our fixations and saccades with black areas in between. Although here we admit that we may be pushing the explanatory power of our theory past its limits, one possible explanation relates to the type of information that our conscious memory system remembers. In general, we tend to remember the general concept, idea, or gist of the information (Reyna and Brainerd, 1995). Because we now understand that we do not directly consciously perceive visual information—we remember it—we may be remembering the gist of the visual scene in the same way that we might remember the gist of a conversation, movie, or list of items. If consciousness does not have the function of capturing our occurrent sensory input, then these features of our visual experiences are not problematic.
Our view of conscious perception is thus consistent with that espoused by Cohen et al (2016). Using an understanding of perception that arises from the field of visual ensembles and summary statistics, they described how observers can extract the gist of the scene with just a few fixations.
Does Conscious Perception Overflow Working Memory?
Block (2011) and others have suggested that phenomenally conscious perception has a rich content with a greater capacity that “overflows” the more limited access consciousness of perception. Evidence supporting this overflow idea comes from the Sperling (1960) paradigm, in which an array of letters (eg, in a 3×4 grid) is briefly shown. Participants generally report seeing all (or almost all) of the letters, yet they can report only three to four letters. Powerfully, however, when they are cued to report the letters from any row, they are able to recall three to four letters in that row, suggesting that all 12 of the letters were potentially accessible. Our memory theory of consciousness would explain that the full array of items is present briefly in unconscious perceptual processes, although we only experience the conscious perception of the items that our attention is drawn to by the poststimulus cue, as only after our attention is drawn to one row of letters are those items transferred into sensory memory and then into working memory.
This explanation is analogous to the one we previously discussed for Sergent and colleagues (2013). Kentridge (2013, p. R71) noted that Sergent and colleagues’ result “does not necessarily invalidate the distinction between access and phenomenal consciousness, but it does lend weight to the alternative, and perhaps simpler, position that consciousness is just consciousness.” We agree with Kentridge (2013) and Cohen et al (2016) that although distinguishing between phenomenal and access consciousness appears to be a useful distinction, it leads to strange situations in which we can have a phenomenally conscious experience that our conscious mind does not have access to. This could be rephrased to say that there can be phenomenally conscious experiences that are unconscious—an idea that does not make much sense to us. We believe that such distinctions have been postulated to solve some of the problems of consciousness—problems that we believe our memory theory of consciousness can explain without the need to invoke phenomenal and access consciousness.
We can now understand why it is difficult to control our thoughts when we practice mindfulness. Our theory is that consciousness did not develop in order for us to perform high-level abstract reasoning using language, logic, visuospatial abilities, or other cognitive capacities in order to allow us to carry out intentional actions. Instead, consciousness developed in order for us to remember events and information, as well as to creatively and flexibly recombine those events and information. We speculate that much of this remembering—and even the recombination of the remembered events and information—can occur without volitional control over consciousness. In other words, we may consciously perceive events and, at a later time, consciously imagine the recombination of elements of those events, even if the recombination is being directed by unconscious processes. As anyone knows who has practiced mindfulness, it can take great effort to sustain conscious attention to a single object—because, we argue, that is not what consciousness developed to do.
If we consider mindfulness from our new perspective, we might imagine the following processes occurring. A thought is generated unconsciously, perhaps of a meeting we are anticipating later today. Our attention is captured by this thought. Depending on the goals of our mindfulness session, we may be content to simply be metacognitively aware of this thought, observing with some detachment the different emotions the thought produces (eg, excitement that the meeting might go well as well as anxiety that it might go poorly). Or, we may attempt to nudge our awareness from the meeting to our breath as we try to consciously attend to air going in and out of our nostrils.
The Subjective Experience of Consciousness
One of the most exciting consequences of our memory theory of consciousness is how it might possibly explain certain aspects of the subjective experience of consciousness.
Stream of Consciousness
We all feel that James’s (1890) metaphor of a stream of consciousness is intuitively correct, with the momentary now where we are standing in the river, past events flowing progressively downstream, and future upstream events that are going to occur rushing toward us. Part of the power of this metaphor is that it is fairly linear. Yet, we know that the brain is processing a massive amount of information in parallel. Why do we experience events serially instead of in the parallel manner that the brain processes them? We would argue that it is because it is a property of our conscious memory system to remember—and thus to consciously experience—events serially in time. Once we decouple real-time sensory input from consciousness, we no longer need the two to be processed by the brain in the same way.
We also believe that our memory theory of consciousness explains, at least on one level, why events are bundled together over time such that they seem continuous. Again, we would argue that it is part of the neural architecture of our conscious memory system to store and retrieve events that are bundled together and time-stamped in a certain order.
In the Cartesian Theater
We sometimes have the intuitive experience that we, as conscious selves, are sitting inside our head—in the proverbial Cartesian theater—peering out at the world through our eyes, as if we are watching a movie. We believe that this feeling is present because we are not experiencing our perceptions directly; we are experiencing a memory of our perceptions. Who is this homunculus sitting in the Cartesian theater? It is our conscious self remembering our perceptions, decisions, and actions. Why did it develop this way? We believe that it developed this way to allow the flexible recombination of prior events and information through imagination, like a movie director moving scenes around on a storyboard.
Do we now have to deal with an infinite regress of homunculi sitting in Cartesian theaters? We would argue not. We have one conscious self. Our conscious self is sitting in the theater, mostly passively, watching memories of experiences. The various parts of our unconscious self are processing information in parallel, sometimes paying attention to what is going on with the conscious self but mostly ignoring it. There is only one conscious homunculus. It stops right there.
Consistent With Higher-order Theories
Higher-order theories of consciousness claim that a mere first-order representation (eg, being presented with a red stimulus) would only lead to consciousness if we are in some way aware of having that experience (ie, being aware that we are seeing red). These theories also claim that conscious experiences involve some type of minimal inner awareness of one’s ongoing mental functioning due to the first-order representation being monitored or represented by a relevant higher-order representation (Brown et al, 2019).
Although our memory theory of consciousness clearly differs from higher-order theories of consciousness in many ways, we believe that our theory is consistent with the idea that we are not conscious of the first-order brain mechanisms that process the presentation of a red stimulus in front of our eyes, and that we only become conscious of this stimulus when (sitting in the Cartesian theater) we experience the memory of its perception. Thus, we believe that our memory theory can explain this intuitive sense that we do not experience first-order representations directly, but only indirectly. Higher-order theories would say that consciousness is experiencing higher-order representations of the first-order representations, whereas our memory theory would say that consciousness is remembering the first-order representations.
Our memory theory of consciousness helps explain why some metaphors resonate with us intuitively. Now that we understand why the Cartesian theater feels natural, we can understand why Plato’s allegory of the cave resonates with us. In a very real sense, we do not perceive the real world directly; we only perceive shadows or reflections of the real world through memory. Similarly, now that we understand how both the feeling of the theater and the serial appearance of reality are created by our conscious memory system, we can understand why the two metaphors of the movie The Matrix are so powerful: the thought that we are not truly experiencing the world directly, and that the world is actually made up of massive parallel processing streams of data.
Horse and Rider
We would like to introduce one additional metaphor here that captures some of how we think about the conscious and unconscious self. Imagine our brain as a horse and rider together. Our unconscious, System 1 self is the horse, which is in control of the moment-to-moment journey we are taking. Just as we do not need to provide detailed instructions to a horse on how to cross a rocky field or jump over a short wall, neither do we need to provide detailed instructions to our unconscious as to how to carry a cup full of hot coffee across the room and down a flight of stairs—we just need to look at the cup and our unconscious self does the rest. Our conscious, System 2 self is the human rider, who is mostly just going for a ride. The rider can, of course, provide either moment-to-moment or more general, overall instructions to the horse, and the horse is usually happy to oblige.
How does this interaction between the conscious rider and the unconscious horse happen? Metaphorically, the rider says a few words, tugs gently on the reins, or squeezes their legs to let the horse know which way the rider wants to go. Approximately 500 ms later, the rider can then sense whether the horse has, indeed, made the desired decision and moved in the desired direction. Of course, there are sometimes conflicts—such as when the horse wants to go down the easy path but the rider wants to travel up the mountain.
How is it that our System 2 rider is able to inform careful, considered decisions? Well, it is sunny out there in the Plains, and so the rider always wears a special pair of sunglasses. These sunglasses have a high-tech video screen built into them. In fact, these sunglasses have no lenses, only the built-in screen—a screen that has a delay of ∼500 ms. In other words, the rider is always perceiving the world by looking at the screen, which shows the visual world a little bit after the fact. But these special sunglasses allow a variety of amazing options: the rider can either (a) look out at the world (via sensory memory) 500 ms after it has passed by; (b) access prior autobiographical episodic memories using an active, creative, memory-building process; (c) access prior semantic information; (d) keep information on the screen so that it can be manipulated in working memory; or (e) use a combination of these features to flexibly, creatively imagine possible future outcomes.
When faced with a difficult problem, the rider can use all of these tools to come up with one or more possible solutions, which the rider then communicates to the horse, and the horse makes the actual decision—which may or may not be the same as what the rider suggested. Note that because the rider is always perceiving the world 500 ms after the fact, the rider depends on the horse to make decisions without the rider’s input whenever quick, System 1 decisions are needed.
Oh, and there is just one other issue about these special sunglasses that needs to be mentioned: The images come from the horse. The horse is in control of the videos that the sunglasses are showing the rider. We can think of this part of the metaphor as explaining the 500-ms delay that the rider experiences when perceiving the world. But the horse is also in charge of whether the rider views the world, prior autobiographical episodic memories, semantic information, working memory, or imagination. As always, the rider can communicate to the horse which images they wish to see, but the horse does not always comply, either because it cannot or because it wishes to show the rider another image. This is why mindfulness is so hard; the horse has a mind of its own.
To summarize this metaphor, we consider the rider (with their sunglasses) to be the conscious memory system—remembering rather than directly perceiving, deciding, and acting. We believe that this conscious memory system was always involved in providing information that could be used by the unconscious brain to make decisions that are informed by past events and information. Through continued evolution, we believe that the conscious memory system developed additional capacities in humans as described above. By contrast, the horse represents all of the unconscious brain processes.
Limitation of These Explanations of the Subjective Experience of Consciousness
We wish to clearly state that we are well aware that our so-called explanations of the subjective experience of consciousness do not even begin to get at the hard problem of consciousness—how a collection of neurons and supporting brain tissue produces subjective experience. We are, however, hopeful that our slightly increased understanding of the phenomenology of subjective experience of consciousness can help other researchers look in the right locations and do the right experiments to tackle the hard problem. Our suggestion to them is to focus on the conscious memory system.
Our memory theory of consciousness makes it easy to understand how individuals with visual apperceptive agnosia and individuals with blindsight can frequently respond correctly to visual tasks despite lacking conscious awareness of visual objects and other stimuli. Although the visual aspects of their conscious memory system are not functioning properly, such that their sensory and working memory are impaired—meaning that perceptions do not enter consciousness—their unconscious self can still perceive stimuli and respond accordingly.
The Causal Role for Consciousness and How Consciousness Contributes to Evolutionary Success
We have made our case that consciousness did not develop to play a direct causal role in decisions and actions, and instead developed as part of the conscious memory system. However, as implied by our horse and rider metaphor, we believe that, in modern human beings—and probably many other animals as well—consciousness is essential to make good decisions and take proper actions. Consciousness enables System 2 (the rider) to use working memory as well as all prior autobiographical and semantic information to inform important decisions. Consciousness is thus tremendously important for evolutionary success: Without consciousness, System 2 decisions could not be made. If we had no consciousness, we could still make decisions, but they would always be fast, System 1 (horse only) decisions. Consciousness allows us to make slow, carefully considered System 2 decisions. We will discuss these issues in more detail later on when we discuss the implications of our memory theory of consciousness.
Consciousness Is not Epiphenomenal
Although we have argued that conscious memory is important for evolution, we have not yet discussed whether the actual subjective experience of consciousness is epiphenomenal. Our belief is that subjective experience is an inherent property of the conscious memory system, and that to say that we can have one without the other would be analogous to saying that we can have molecular motion without heat. Just as ½MV2 = KT, we believe that use of the conscious memory system produces subjective experience. As mentioned earlier, we do not have the answer to this hard problem of consciousness, but we are hopeful that our theory will move the field toward finding that answer.
NEUROANATOMICAL CORRELATES AND DISORDERS OF CONSCIOUSNESS
As might be expected from our experience in philosophy, experimental psychology, cognitive neuroscience, and neurology, our hypothesis of the neuroanatomical correlates of consciousness relates to brain regions and structures and not to underlying cells, cellular assemblies, or neural oscillations (eg, Lou et al, 2017). We understand that cellular and molecular microstructures may be crucial to gaining a full understanding of the neurophysiologic basis of consciousness, including the hard problem. Again, we hope that this discussion of what we consider to be key brain regions and structures will help others to dive deeper and achieve a more complete understanding of the hard problem. Because much of our discussion of the possible neuroanatomical correlates of consciousness is related to individuals with various brain disorders, we will consider these two topics together.
Other Theories of the Neuroanatomical Correlates of Consciousness
There are currently four major theories that make predictions regarding the neural correlates of consciousness: recurrent processing theory (Lamme, 2015, 2018), global neuronal workspace theory (Mashour et al, 2020), integrated information theory (Tononi et al, 2016), and higher-order thought theory (Brown et al, 2019). Some of these theories specifically address the hard problem, whereas others, like ours, only try to point the way toward possible solutions. As pointed out by Yaron et al (2022), each of these four theories emphasizes largely different cortical regions as being critical for consciousness.
Recurrent Processing Theory
Recurrent processing theory suggests that conscious processing depends on horizontal connections and recurrent loops between lower- and higher-brain regions that are extended in time and space and involve changes mediated by NMDA-dependent feedback activations (Lamme, 2015). Posterior cortical regions associated with visual processing of information are emphasized, with the prefrontal cortex contributing to, but not essential for, conscious processing (Lamme, 2018).
Global Neuronal Workspace Theory
Arising out of the global workspace theory (Baars, 2005; described earlier), global neuronal workspace theory proposes that conscious processes occur when information in specialized processors enters a large-scale reverberant brain-scale network of high-level cortical areas linked by long-distance re-entrant loops and becomes ignited, allowing global access by other specialized processors (Mashour et al, 2020). The parietal and prefrontal cortical areas are critical for routing information between other cortical processors. At the neuronal level, large pyramidal cells in cortical layers II/III and V play key roles in the neuroanatomical correlates of consciousness.
Integrated Information Theory
Integrated information theory attempts to directly address the hard problem by starting from the essential phenomenal properties of experience and infers postulates about the characteristics that are required of its physical substrate (Tononi et al, 2016). It also provides a mathematical quantity of integrated information that yields a measure of the degree of consciousness of any system. The occipital and parietal lobes are considered to be critical and sufficient for conscious experience.
As introduced earlier, higher-order theories postulate that a first-order representation, such as awareness of a rose, would not be sufficient for conscious experience to arise. An organism must be in some way aware of itself as being in a first-order state in order to be conscious of it. Proponents of higher-order theories consider the prefrontal cortex to be important for conscious perception (Brown et al, 2019).
Neuroanatomical Correlates of the Conscious Memory System
Having reviewed some of the leading theories that suggest which brain regions are important for consciousness, we will now state our hypothesis, which we will then work to support. Because we contend that the conscious memory system supports both consciousness and all forms of explicit memory, we hypothesize that the neuroanatomical correlates of consciousness are the neuroanatomical structures that are involved in all forms of explicit memory.
Which structures are involved in explicit memory? The hippocampus should certainly be included along with related structures, such as the neighboring entorhinal and perirhinal cortex, as well as Papez’s circuit, including the fornix, mamillary bodies, and anterior nucleus of the thalamus. We would also argue that the cerebral cortex is necessary for explicit memory. We might immediately think of the inferolateral temporal cortex for semantic memory, and then perhaps the frontal and parietal cortex for working memory, including regions that are involved with the default mode network, such as medial prefrontal cortex, posterior cingulate cortex, precuneus, and angular gyrus (eg, Zheng et al, 2021). But, we take the view espoused by Murray et al (2020, p. 2) in their book, The Evolutionary Road to Human Memory. They argue that the entire cerebral cortex is important for memory, writing:
In our opinion, every cortical area contributes to memory, each in a specialized way. As our ancestors traveled along their evolutionary trajectories, cortical areas accumulated over time; and, in each instance, this happened for the same fundamental reason: to transcend problems and exploit opportunities that these animals faced in their time and place.
The Evolutionary Road to Human Memory (Murray et al, 2020) provides a wonderful summary of the animal, human brain lesion, and neuroimaging studies that support the view that all cortical structures are not only involved with, but are critical for, a specific type of memory that is needed for a specific type of task, whether it be navigating the wilderness, distinguishing the sounds of prey and predators, or recognizing faces.
We contend not only that every cortical region contributes to memory, but that they each also contribute to a specific domain of conscious awareness. For example, we believe that the following speculations are likely correct:
- Visual areas in the occipital cortex are necessary for visual consciousness, including visual imagery.
- The auditory cortex in the superior temporal cortex is necessary for auditory consciousness.
- The parietal cortex (particularly of the nondominant hemisphere) is necessary for conscious awareness of space (particularly on the contralateral side).
- The primary somatosensory cortex in the postcentral gyrus of the parietal lobe is necessary for certain subdomains of sensory consciousness such as graphesthesia (identification of numbers or letters written on the skin) and stereognosis (identification of objects by touch).
- The primary motor cortex in the precentral gyrus of the frontal lobe is necessary for conscious awareness of the fine motor movements that are used to play musical instruments, thread needles, and perform other delicate tasks.
- The frontal eye fields are necessary for conscious awareness of eye movements.
- Broca’s area is necessary for conscious awareness of our own speech, whether vocalized or just “in our head.”
- The insular cortex is necessary for conscious awareness of our body.
- The prefrontal cortex areas that facilitate complex thought, working memory, problem-solving, and judgment are necessary for the conscious awareness that comes with these higher-level abilities.
The work of Gazzaniga (2015) and others with split-brain patients has demonstrated that when the corpus callosum is severed, the individual may be left with two separate consciousnesses in one brain that, seemingly, work together just fine without any apparent functional difficulty or conflict (or, at least, no more internal conflict than any of us have with our unsplit brain from time-to-time). We believe that something analogous occurs not only with the left and right hemispheres, but also with each region of the cortex.
We further suggest that each region of the cortex is autonomously conscious, and its island of consciousness is not dependent on any central executive or other region. That is, the minimally sufficient cortical region needed for conscious awareness may be any cortical region, whether it be a sensory region enabling conscious perception, a motor region enabling conscious movement, or an association region enabling some type of multimodal thought. We believe that this must be the case because, as far as we are aware, there is no single cortical region (unilateral or bilateral) that, when removed, renders the individual unconscious. Thus, we contend that the visual conscious awareness in the occipital cortex is independent of the auditory consciousness in the superior temporal cortex, and both are independent of the conscious awareness and guidance of motor movements occurring in the frontal cortex.
What is the evidence for this hypothesis that any cortical region may be autonomously conscious? In large part, it comes from the experience of working with several thousand patients who have had strokes or neurodegenerative diseases affecting every part of their cortex, and is supported by the human patient literature. It is well known that damage to subcortical structures that are part of the reticular activating system (such as portions of the midbrain and thalamus) can lead to unconsciousness (eg, Kinney et al, 1994), but, again, we know of no cortical regions that lead to unconsciousness when they are damaged. Indeed, it might be that some of these subcortical reticular activating system structures (such as the thalami) act as a hub, switching between different conscious cortical regions.
The literature supporting our view that (a) disruption of no cortical region (even widespread frontal or occipital/parietal regions) leads to unconsciousness but (b) virtually all cortical regions contribute to consciousness has been well reviewed by the opposing front versus back cortical theories of consciousness (Boly et al, 2017; Michel and Morales, 2020; Odegaard et al, 2017). We believe that our theory can reconcile those views espoused by the opposing front versus back camps.
For example, if posterior brain regions are critical for consciousness, as suggested by recurrent processing theory and integrated information theory, how is it that patients with posterior cortical atrophy or bilateral occipital and parietal strokes are not unconscious? These patients most certainly have deficits of conscious awareness, but we would never say that they are unconscious. Similarly, if the prefrontal cortex is critical for consciousness, as suggested by global neuronal workspace theory and higher-order theories, what about patients with behavioral variant frontotemporal dementia or bilateral frontal strokes? There is no doubt that the consciousness in these patients is impaired, but we would not say that they are the physical manifestation of unconscious philosophical zombies appearing to have a conscious inner world when they actually have none. We believe that they could consciously experience the beauty of a sunset as well as anyone else. Which group of patients is not conscious (in addition to those with reticular activating system damage)? We believe that it is individuals with either widespread or diffuse cortical dysfunction. These awake-but-not-conscious individuals are described as having an encephalopathy or delirium.
Exactly how large would a cortical region need to be in order to support some form of independent consciousness? We do not know the answer, but we will suggest some methods to address that question in Future Directions.
Note, however, that we are not saying that the only function of the cortex is to provide conscious awareness for specific modalities. The cortex certainly performs much additional work that is solely involved in the unconscious processing of information.
We will also point out that saying that the entire cortex is involved in consciousness is not the same as saying that the entire brain is involved in consciousness. For those readers who think of the cortex as being more or less synonymous with the brain, please note that although the cerebral cortex comprises 82% of the mass of the human brain, it contains only 16 billion (19%) of the 86 billion neurons that are in the brain (Herculano-Houzel, 2009).
We now turn to a brief review of some of the major brain regions and structures that are involved in the conscious memory system, along with some of the relevant neurologic disorders that impair consciousness in one way or another.
Hippocampus, Related Structures, and Individuals With Amnesia
When considering the hippocampus, we typically also consider both neighboring medial temporal lobe structures such as the entorhinal and perirhinal cortex, as well as anatomically connected structures such as the fornix, mamillary bodies, and anterior nucleus of the thalamus. Whether because of neurodegenerative disease, infection, inflammation, stroke, seizure, or surgery, the cognitive effects of damage to the hippocampus and related structures are well known. Disruption of episodic memory invariably occurs, leading to anterograde and some retrograde amnesia. Consciousness, at least in the ordinary sense of the term, does not appear to be disrupted by damage to the hippocampus and related structures. Individuals with such damage can certainly consciously experience many perceptions, decisions, and actions.
However, individuals with damage to the hippocampus and related structures do show impairment in several aspects of cognition that we would argue are related to consciousness. First, individuals with hippocampal damage typically lose the ability to consciously perceive subtle differences in visual scenes, and individuals with neighboring perirhinal cortex damage typically lose the ability to consciously perceive subtle differences in faces (Mundy et al, 2013). Second, these individuals show reduced ability to imagine the future (Addis et al, 2009) and therefore to plan for it. Third, by losing the ability to form new episodic memories, these individuals are impaired in their ability to update their sense of self. We consider the ability to imagine the future and to update one’s sense of self to be important aspects of consciousness.
Lastly, we speculate that consciousness would almost certainly be impaired if the hippocampus and related structures were completely absent from birth. Although studies have shown that individuals with perinatal damage to some regions of the hippocampus develop normal semantic memory and do not show any obvious impairments in consciousness (Elward and Vargha-Khadem, 2018), we believe that consciousness would not be normal in individuals with complete absence or complete dysfunction of the hippocampus and related structures since birth.
Occipital Cortex, Visual System, Anton Syndrome, Blindsight
We described previously how individuals with damage to the occipital cortex leading to visual apperceptive agnosia or hemifield blindness can sometimes perform tasks unconsciously that depend on vision despite the fact that they cannot perform the task with conscious awareness. We believe that these cases support the idea that it is the visual cortex in the occipital lobes that provides conscious awareness of vision.
Individuals with damage to their entire occipital cortex bilaterally, leading to complete cortical blindness, may deny that they are blind (Anton or Anton-Babinski syndrome), which is a form of anosognosia or unawareness of their deficit (Das and Naqvi, 2022). We believe that this syndrome is one example of a general phenomenon whereby the brain region that generates a specific aspect of consciousness—in this case, visual perception and visual imagery—is the same part that is responsible for the awareness of whether that aspect of consciousness is present, absent, or distorted. Our prediction would be that Anton syndrome would not be present if the damage to the visual system affected pathways before the occipital lobes (such as the eyes, optic nerves, tracts, radiations, or lateral geniculate nuclei). Thus, we believe that the phenomenon of Anton syndrome supports the idea of visual consciousness being present in the occipital cortex.
Parietal Cortex, Spatial Neglect, and the “Aha” Moment of Recollection
The parietal lobe plays an important role in attention to, or awareness of, one side of the world. Although both parietal lobes contribute to awareness of both the left and the right side, damage to the parietal lobe of the language-dominant hemisphere (usually left) typically produces a mild and temporary loss of awareness (or neglect) of the contralateral right side, whereas damage to the nondominant (usually right) parietal lobe typically produces prominent and sometimes permanent neglect of the contralateral left side (Mesulam, 1999).
Neglect most commonly occurs for sensory stimuli that are localized in space, such as visual and tactile stimuli. Individuals with neglect may not notice food on the left half of their dinner plate, may find it difficult to pay attention to someone who is speaking to them on their left, and may not notice a touch to the left side of their body.
What is particularly striking about individuals with right parietal lobe damage and left neglect is the fact that they are typically completely unaware that the left side of the world is missing. As in Anton syndrome, individuals with neglect have anosognosia and are unaware of their deficit. We believe that parietal damage-induced neglect supports the idea of spatial awareness (consciousness) of one side of the world being present in the parietal cortex, particularly of the nondominant hemisphere.
Other parietal lobe functions may also be relevant for consciousness. With the exception of the retrosplenial cortex, damage to the parietal lobes is not known to impair episodic memory. Yet virtually every recognition memory task evaluated with fMRI or event-related potentials (ERPs) produces activation of the parietal cortex (Simons et al, 2008). How are we to reconcile this discrepancy? Our answer is that the so-called parietal old–new effect (in which previously seen or old items show greater parietal activation compared with novel or new items) is part of the neural correlate of the conscious awareness that an item has been seen before. It is that aha moment when we consciously think, “Yes, I remember that” (Ally et al, 2008).
Frontal Cortex, Motor System, and Corticobasal Syndrome
Regarding consciousness and the frontal cortex, we begin by noting that neglect also occurs for movements and activities after damage to the supplementary motor cortex (Laplane and Degos, 1983). Individuals with this type of damage might abandon washing, shaving, or brushing their hair on one side. Similarly, individuals with damage to the frontal eye fields typically show a form of neglect in that they experience difficulty moving their eyes voluntarily to the opposite side. We believe that these motor forms of neglect support the idea that the frontal lobes are important for the conscious control of movements and activities.
Some individuals with brain lesions experience a bizarre delusion in which they do not recognize or believe that their paralyzed limb cannot move (ie, anosognosia) or even that it is part of their body (ie, somatoparaphrenia). Brain regions implicated in somatoparaphrenia include portions of the insula and the middle and inferior frontal gyri (Gandola et al, 2012). Thus, conscious awareness of movement, and even awareness that a body part is one’s own, may be related to proper cortical functioning.
Some individuals with corticobasal syndrome experience alien limb phenomena, in which an affected limb appears to move on its own—without conscious control. We have seen patients in which the usually useless limb may rise and make simple movements. When the patient is performing tasks that would typically require two hands, such as tying shoelaces, the useless limb can sometimes be cajoled into helping, which it then may do easily and automatically (A.E.B. observation). Although individuals with corticobasal syndrome show involvement in the basal ganglia in addition to the cortex, atrophy and/or hypometabolism of the frontal and parietal cortex is prominent and generally detectable on brain imaging studies. This phenomenon supports the idea of the importance of the cortex in conscious motor control.
Individuals with behavioral variant frontotemporal dementia show atrophy and/or hypometabolism of various regions of the frontal cortex, leading to behavioral problems. Many of these individuals demonstrate utilization behavior, which occurs when individuals see a tool and automatically start to use it. An individual may pick up a pair of scissors and begin cutting or a pen and begin writing. We view utilization behavior as an example of damage to the frontal cortex, leading to impaired conscious control of behavior but preserved unconscious actions. There are many other examples of damage to the frontal lobes leading to a loss of conscious control of behavior (eg, Phineas Gage [Damasio et al, 1994]), which we will not review here.
Apathy is a very common symptom in individuals with a variety of damage to the prefrontal cortex. Here, we would argue that some of the so-called higher frontal lobe functions, such as planning and abstract reasoning, are impaired because conscious awareness and control of such functions has been impaired. Not consciously realizing that one should be planning and acting leads to apathy.
Temporal Lobes, Auditory Cortex, Inferolateral Temporal Cortex, Names, Words, and Meaning
Important aspects of the temporal lobe cortex include parts of the insula, the auditory cortex, and the vast store of information that comprises semantic memory. Individuals with damage to the bilateral auditory cortex lose conscious awareness of sounds but still may react to them, which is a phenomenon that is considered analogous to blindsight (Cavinato et al, 2012). Animal studies have shown experimentally that damage to the bilateral auditory cortex does not necessarily change behaviors in response to sounds (Floody et al, 2010). Thus, we have evidence that the auditory cortex is involved in the conscious awareness of sounds in a manner that is consistent with our hypothesis.
Evidence from both imaging and individuals with brain lesions suggests that the inferolateral temporal cortex is critical for conscious awareness of the names of people, animals, and tools (Damasio et al, 1996). Moreover, although unilateral left-sided damage may impair access to just the names of such items, bilateral damage may lead to a complete loss of knowledge of animals, plants, and man-made objects. Individuals with neurodegenerative disease, including semantic dementia and Alzheimer disease (AD), frequently experience this loss of knowledge in the later stages—not knowing what a rabbit, pumpkin, or remote control is—which, we would argue, is the loss of a form of awareness of, or consciousness of, the items. In addition, even very early on in the course of their disease, once the concept of an item is completely lost from semantic memory, individuals with these semantic memory deficits are not consciously aware that anything is wrong—they just believe that they may never have encountered the missing items and therefore have no knowledge of them.
Properties of the Cortex That Support Consciousness
We hope that we have made our case that each cortical area contributes to specific conscious awareness related to the function of that cortical area, and that destruction of any cortical area will disrupt or abolish the domain of consciousness that was supported by that area without disrupting other domains of consciousness. Our hypothesis that all cortical regions contribute to consciousness is consistent with a recent neuroimaging study that compared conscious versus unconscious object recognition and found widespread areas of cortical involvement during conscious object recognition (Levinson et al, 2021).
Our anatomical hypothesis also provides one way to reconcile the different predictions made by the four major theories of consciousness regarding which cortical regions are most important (see Yaron et al, 2022, for review), arguing that they all are, but none is critical. We are not implying that subcortical structures are not necessary for consciousness—they most certainly are—but they are not sufficient for consciousness. Furthermore, subcortical structures do not contain the necessary neuronal architecture that allows for phenomenology or qualia of consciousness to occur.
Although we consider cerebral cortex architecture as the unit of neuronal assemblies that allows consciousness to occur, exactly how consciousness arises from the cortex is unclear. We will leave it to others to determine whether consciousness is related to the spindle neurons (also known as von Economo neurons) found in layer V, thick-tufted pyramidal neurons in layer VB, large pyramidal cells in cortical layer II/III, other neuronal types, or none of these. It may be that assemblies of neurons are the unit of consciousness, and that looking at single cells in order to understand consciousness is similar to studying the nature of quarks inside the neutrons of iron atoms in order to understand why iron filings are attracted to magnets. Thus, it may be that concepts such as the perturbation complexity index (PCI) that evaluate the integrity of brain networks will be key to understanding consciousness (Koch et al, 2016).
Supposing that assemblies of neurons in the cortex are the unit of consciousness, how many layers of the cortex are required? Is only a 6-layered cortex conscious, or is a 5-, 4-, or 3-layered cortex sufficient? Can a 3-layered allocortex produce some “low-level” perceptual or emotional consciousness, whereas a 6-layered neocortex is required for higher-level self-consciousness and abstract reasoning? Because our memory theory of consciousness is that consciousness developed as part of the conscious memory system, which includes hippocampally based episodic memory, we speculate that some conscious awareness is present in even the allocortex.
Whatever the correct answer, we would argue that our lack of understanding of exactly how the cortex produces consciousness does not prevent us from using this hypothesis of the involvement of the cortex in consciousness to discuss some implications, which we will do now.
Our memory theory of consciousness, with regions of the cerebral cortex as the fundamental units that allow for conscious awareness, leads to a number of implications regarding which animals are conscious, which neurologic and psychiatric disorders may impair consciousness, and how the conscious mind and unconscious brain work together day by day, minute by minute, and second by second. These implications, in turn, raise ethical implications, which we will also briefly discuss.
In the prior section, we speculated that conscious awareness of basic perceptions and emotions may be present with a 3- or 4-layered allocortex, whereas conscious abilities such as problem-solving may require a 6-layered neocortex.
Consciousness in Mammals
Because all mammals have a neocortex (Kaas, 2019), we argue that all mammals are conscious. Paralleling their differences in cortical structure, we believe that the consciousness of mice differs in its complexity from that of a dog, which differs from that of a chimpanzee, which differs from that of a human. For example, Pine et al (2021, p. 701) reviewed some of the changes in the neuroanatomy between nonhuman and human primates, such as the expansion of homotypical association areas and the hippocampus and how these expansions are related to “(i) a subjective sense of participating in and re-experiencing remembered events; and (ii) a limitless capacity to imagine details of future events.” Although we certainly agree that humans developed these capacities to a greater extent than all other species, we also believe that all mammals have some conscious awareness of (ie, memory for) perceptions, decisions, and actions. See Carruthers (2015) for a similar argument that is also based on neuroanatomical homology.
Does this mean that all mammals have human-like conscious memory abilities to mentally time-travel? Suddendorf and Corballis (2007, p. 307) argued that there is little evidence to suggest that nonhuman mammals have developed this mental time-traveling ability, at least in the way that humans do, stating, “We maintain that the data so far continue to suggest that mental time travel is unique to humans.” Although we do not contest their statement, we would still argue that nonhuman mammals possess some form of conscious memory that allows them to have some conscious awareness that provides them with evolutionary advantages, such as being able to determine temporal order (Eichenbaum et al, 2005).
What about the mirror self-recognition test? In this test, animals have a spot of color applied to their head when they are anesthetized or otherwise unaware of its application, in a location that they can only see in a mirror, such as their forehead. If the animal looks in the mirror and reaches for the spot or tries to rub the spot off, then we know that the animal recognizes itself in the mirror. If not, we presume that the animal does not realize that it is its self in the mirror.
Although a number of mammals have been shown to pass this test, including the four great apes, bottlenose dolphins, and Asian elephants (de Waal, 2019), most other mammals have not, including our feline and canine companions (although dogs can pass an odor version of the test [Horowitz, 2017]). We believe that this test tells us something about visual perception and the recognition of self-consciousness, but we would argue that just because an animal fails the mirror test does not mean that it has no conscious awareness of any type.
Consciousness in Nonmammalian Species
What about consciousness in other vertebrates such as birds, lizards, amphibians, and fish? Magpies have, in fact, been shown to pass the mirror self-recognition test (Prior et al, 2008), as has a species of fish (Kohda et al, 2019). Does this mean that certain bird and fish species are self-conscious? de Waal (2019) argued that passing the mirror test indicates at least some rudimentary self-awareness; however, one researcher suggested that we should consider all the data that shed light on the cognitive capacities of a species before drawing conclusions on its self-awareness (or lack thereof) (Vonk, 2020).
Our theory would be aligned with de Waal’s (2019), that consciousness and self-awareness among species are on a continuum. Based on both experimental work and the brain anatomy of vertebrates, we believe that there is evidence that most vertebrates have at least some rudimentary conscious memory system because they have some form of a hippocampus (or a brain structure analogous to it) and some cortex (or analogous structure). It follows logically that those vertebrates that have the requisite anatomy for the conscious memory system would experience at least some conscious awareness, although it might be little more than perceptions and/or emotions. Some researchers have, in fact, argued that at least one nonmammalian vertebrate species, the California scrub-jay, can use its memory to spontaneously plan for the future without reference to its current motivation—something that had previously been thought to be a uniquely human ability (Raby et al, 2007; see also Carruthers, 2015).
Our theory is also consistent with many of the ideas put forth by Ginsburg and Jablonka (2021), who go even further than we do. They suggested that “unlimited associative learning” can be considered a marker of minimal consciousness. They noted that such learning is present not only in almost all vertebrates, but also in octopods, squid, cuttlefish, honeybees, and cockroaches.
Ethical Implications of Conscious Vertebrates
We are not vegetarians and do not want to imply that the inevitable ethical conclusions that stem from our theory mean that we should all become vegetarians (or at least, nonvertebrate consumers) in order to avoid harming conscious animals. However, we would argue that we as societies and individuals should consider the ethical implications of consuming cows, pigs, chickens, and other vertebrates that we argue have forms of conscious awareness.
Disorders of Consciousness
Now that we have a better understanding of both the phenomenology of consciousness and the neuroanatomical structures needed for it, we can speculate that a number of psychiatric, neurologic, and developmental brain disorders may be disorders of consciousness.
We reviewed earlier how strokes that damage certain areas of the cerebral cortex result in specific impairments in consciousness that are related to the functions of those particular areas. Here, we will simply add that strokes in the subcortical white matter of the corona radiata may also cause impairments in consciousness by disconnecting cortical regions from one another, leading to, for example, Wernicke aphasia (Mesulam et al, 2015) or alexia without agraphia (Geschwind, 1965). In brief, strokes that affect cortical and/or subcortical white matter frequently impair one or more domains of consciousness, the conscious memory system, and the ability to use previously learned information to problem-solve and plan for the future.
Individuals with delirium (also known as an acute confusional state or encephalopathy) show an inability to carry out a coherent stream of thought or action (Lipowski, 1989). We suggest that these individuals manifest a primary disorder of consciousness. As expected, they are also unable to properly store, retrieve, or use conscious memory for goal-directed activity.
Individuals with delirium can (and often do) engage in a variety of activities, including running, undressing, eating, and performing other complex activities (Lipowski, 1989). We suggest that such activity is performed without conscious awareness or control, and it is because of this lack of conscious awareness and control that the activities result in no episodic memories being formed and essentially no complex goals being achieved. (Individuals may undress because they are warm or drink because they are thirsty, but these types of basic goals are the only ones that can be achieved.)
Delirium may result from cortical dysfunction caused by metabolic disturbances (such as elevated calcium) or the inflammatory processes that accompany systemic infections (such as pneumonia) (Wilson et al, 2020). Delirium is more common when the cerebral cortex has already been damaged by strokes or degenerative diseases (such as AD).
AD and Other Cortical Dementias
We contend that whether or not all cortical dementias start with involvement of the subcortical structures (such as the substantia nigra of the midbrain in Parkinson disease dementia), these dementias eventually impair one or more domains of consciousness due to their involvement of the cortical regions. Cortical dementias include AD; Lewy body dementia (including Parkinson disease dementia and dementia with Lewy bodies); progressive supranuclear palsy; corticobasal degeneration; and the frontotemporal lobar degenerations that may lead to behavioral variant frontotemporal dementia, primary progressive aphasia, or progressive amnestic dysfunction, depending on where in the cortex the pathology starts.
We believe that cortical dementias impair different domains of consciousness depending on which cortical regions they disrupt, just as strokes do. In addition, the pathology in cortical dementias spreads throughout the brain, often affecting both adjacent cortical regions and regions that are neuronally connected. For example, AD’s pathology spreads throughout the default mode network, which consists of the hippocampus; related medial temporal lobe structures; anterior and lateral temporal cortex; medial prefrontal cortex; and posterior cingulate, precuneus, and angular gyrus in parietal lobes (Buckner et al, 2008).
As cortical dementias spread and the pathology involves more than one cortical region, we believe that more than one domain of consciousness becomes impaired. Once several cortical regions are involved, a phenomenon known as sundowning becomes more likely to occur. Most cases of sundowning are a temporary and periodic state of delirium. Over time, most cortical dementias spread to most cortical areas of the brain, which we believe causes impairment in most domains of consciousness. Ultimately, the individual with cortical dementia spends more time in a delirious, confused—and, we would argue, unconscious—state, with few periods of conscious lucidity.
Individuals with classic migraines experience auras that are most commonly visual but may also cause numbness, speech or language difficulties, weakness, or confusion. Although migraine auras are not completely understood, it is clear that they are associated with cortical spreading depression (also known as spreading depolarization), in which there is a spreading of electrophysiological hyperactivity over the cortex at a velocity of several millimeters per minute followed by spreading inhibition. Phenomenologically, the most common experience is that of positive visual phenomena (often described as scintillating zigzag lines or fortifications) followed by a lack of vision or scotoma. Both the scintillations and the scotoma progress over minutes and generally resolve in under an hour. Relevant here is the fact that cortical spreading depression causes disturbances of conscious perception and sometimes other conscious abilities.
Similar to individuals with migraine auras, individuals with epilepsy may consciously experience an “aura” of positive phenomena, such as flashes of light or tingling and numbness, which usually spreads over seconds. Following the positive phenomenon, a postictal loss of function may occur (sometimes called a Todd paresis when it is related to motor function in a limb). In addition to visual and tactile, auras may also be olfactory or gustatory, or they may cause simple or complex motor movements (such as lip smacking or bicycling movements). Seizure auras have been shown to be focal seizures that occur in some individuals before the seizure becomes generalized. Once the seizure becomes generalized, the individual loses consciousness. Again, relevant here is that the temporary disruption of cortical activity may cause focal or general disturbances of consciousness, depending on the extent of the cortical disruption.
Individuals with dissociative identity disorder have two or more distinct and relatively enduring personality states (American Psychiatric Association, 2013). This disorder generally develops in childhood due to prolonged and severe physical, sexual, and/or emotional abuse. We speculate that individuals with this disorder may have two or more separate consciousnesses, each with its own memories, that the other consciousnesses may or may not have access to. If we are correct, it may be an example in which psychological factors can cause a disorder of consciousness.
There is evidence from the literature, such as cases reported by James (1890), that not only can episodic memories be separated between identities, but semantic memories can as well. In one case, James (1890) described that when the individual moved from one identity to another, her prior knowledge base was not accessible and she had to learn (or relearn) information that her other self knew.
Another related disorder is dissociative amnesia, in which an experienced or witnessed traumatic event—so emotionally laden that it would typically not be forgotten—cannot be remembered. We believe that this disorder may be considered a more focal case of the conscious memory system being fractured such that one or more explicit memories cannot be retrieved.
Depersonalization–derealization disorder includes symptoms of feelings of detachment from one’s mental or bodily processes or one’s surroundings. It is thought to be caused by early childhood emotional abuse and neglect. Triggers of depersonalization symptoms may include significant stress, panic attacks, and drug use. We speculate that, at least in some cases, this disorder may be a manifestation of a disruption of consciousness.
Schizophrenia consists of positive symptoms of psychosis, such as hallucinations, delusions, and disorganized thinking, and negative symptoms of social withdrawal, decreased emotional expression, and apathy. Although all modern theories of schizophrenia state that it is a disorder of the brain resulting from interactions between individuals’ genetic makeup and their environment, the specific biological pathophysiology has not been determined. We believe that schizophrenia may represent a disorder of consciousness, with the hallucinations (usually of voices) being a symptom of disrupted and possibly fragmented consciousness (Tordjman et al, 2019). This idea is supported by the observation that individuals with schizophrenia show impaired conscious access but intact unconscious processing of perceptual stimuli (Berkovitch et al, 2017). Consistent with our memory theory of consciousness, memory deficits are common in individuals with this disorder (Avery et al, 2020).
The autism spectrum includes individuals of highly varied capacities, from those who are nonverbal and function at a cognitive level of <2 years, to those who would have previously been diagnosed with Asperger syndrome and may show superior intellectual abilities. For the purposes of this paper, our brief discussion of autism will be restricted to individuals with high needs, focusing on nonverbal individuals with significant impairment and intellectual abilities less than those of a typical 2-year-old child.
Autism may be caused by several underlying brain differences, leading to several different phenotypes. In individuals with severe classic Kanner autism, there appear to be what we would consider severe deficits in the conscious memory system. Regarding memory, these individuals tend to learn in stereotyped ways, not in the typical episodic-memory-leading-to-semantic-memory manner. Education of these individuals therefore often involves operant conditioning (eg, applied behavior analysis) to produce learning.
Problem-solving and responses to conflict and discomfort appear to us to predominantly use System 1 unconscious processes. There is little evidence (at least to the observer) for conscious, thoughtful, deliberative System 2 processing in some of these individuals. The unfortunate result for many is that they learn, via operant conditioning, maladaptive behaviors such as aggression and self-injury because those behaviors have previously provided them with an escape from activities that they did not like (Oliver and Richards, 2015).
We therefore speculate that individuals with severe classic Kanner autism have a disorder of the conscious memory system that manifests as impairments in both memory and consciousness. This speculation is supported by pathology studies showing that the cerebral cortical architecture is disrupted in these individuals (Courchesne et al, 2011). Other clinicians and researchers have also speculated that autism may be related to deficits in consciousness (Tordjman et al, 2019). However, this speculation should not be misconstrued as suggesting that autistic people are somehow permanently unconscious or incapable of human experience.
There is also interesting anecdotal evidence that some individuals with Kanner autism appear to be able to process information in parallel fashion more easily than typical individuals; for example, some desire to listen to music and watch an unrelated movie at the same time (A.E.B. observation). This comfort with parallel processing may be related to a lack of serial one-thing-at-a-time thinking that is imposed by consciousness.
Lastly, we will note that just as no two neurotypical individuals are the same in their capacities, neither are two individuals with severe classic Kanner autism, such that this discussion may seem consistent with the capacities and limitations of some individuals and inconsistent with the capacities and limitations of others.
Most of our Decisions and Actions Are Unconscious, Although we Think They Are Otherwise
One of the most radical implications of our memory theory of consciousness is that although we generally have the subjective experience that most of our decisions and actions are consciously controlled and few occur automatically and unconsciously, we argue that it is the other way around—few of our decisions and actions are consciously controlled, and most occur automatically and unconsciously.
Let’s take our morning routine, for example. We believe that we are acting primarily unconsciously as we get out of bed, walk to the bathroom, flip on the light, use the toilet, brush our teeth (including all the steps such as wet our toothbrush, put on toothpaste, move the brush across our teeth, rinse our toothbrush and mouth, put the brush back), shower (including all the separate steps), dry ourselves, dress, fix our hair, and so on. It would only be if something were unusual—such as we needed to dress for a job interview or a snowstorm—that there would be much conscious deliberation.
In fact, not only did we not make conscious decisions to guide our actions, but we likely were also only partially aware of our morning routine. There may have been moments when we attended to what we were doing, allowing us to form episodic memories of those moments. At other times, however, our mind was free to use our conscious memory system to review what happened yesterday, reminisce about a prior event, or imagine what will likely happen in our 9 AM meeting. If these latter thoughts were occurring during the time that we were showering, instead of remembering the steps of our shower, we would remember the topics that we were consciously reviewing, reminiscing, and imagining. But, even if we were perceptually aware of each step of our shower, that does not mean that we were consciously controlling those steps; our awareness simply allowed us to remember each step.
Conscious or Unconscious Decision?
It may not be controversial to say that we typically go through our morning routine unconsciously, and that we would only remember whatever we were paying attention to at the time. But, now we arrive at work, drink some coffee, and walk to our meeting. We are a little nervous because we need to explain to our boss why our team did not meet its quarterly target, and we are going to have to explain that it is because our colleague, Biff, who is also in the conference room, did not do his part. To our surprise, as we launch into our explanation, we hear ourselves taking responsibility for the team’s shortfall and not blaming Biff. As we are leaving the meeting, we consider what just happened. We end up consciously thinking that we said what we said because it probably was not really Biff’s fault, mistakes can happen to anyone, and, after all, we were leading this team, so we were ultimately responsible. We nod our head as we think, “Yes, that’s what happened,” and turn to other work.
However, we speculate that what actually occurred is that as we started to speak, we were also unconsciously processing in parallel the facial expressions, body language, and eye movements of everyone in the room—particularly those of Biff and our boss. Our unconscious perceptions of Biff's slightly red and determined face led us to unconsciously realize that he was not going to quietly accept that it was his fault; he would likely deny responsibility and blame us instead. Our boss’s body language—mimicking Biff’s posture and leaning toward him slightly, signifying that our boss was going to side with Biff—was also unconsciously processed. Our unconscious System 1 quickly made the decision to change tactics and say it was our fault—regardless of the truth or of what our conscious System 2 had planned to say. (Remember that we believe that unconscious brain processes make the final decisions—the rider can indicate to the horse where they wish to go, but in the end, the horse decides.)
This scenario raises the questions of (a) why we did not realize what actually happened in the conference room and (b) why we came up with an alternative not-quite-right explanation. We argue that the answer to the first question is that we did not realize what happened because we were not consciously aware of it at the time, and thus we could not consciously consider it then nor could we use our episodic memory (part of the conscious memory system) to remember it. (We can, of course, train ourselves to notice subtle facial expressions, body language, and eye movements, which we could then be consciously aware of and remember.) The second question can best be answered by considering an experiment that was performed by Gazzaniga (2015) and his colleagues with one of his split-brain patients.
The Conscious Interpreter
An individual who had undergone a callosotomy to separate his left and right hemispheres in order to reduce the frequency of disabling seizures was shown separate images in his left and right visual fields. The individual’s task was to use each hand to point to a card that corresponded to the visual image he saw in each visual field. His right visual field, projecting to the verbal left hemisphere, was shown a chicken foot, and his right hand (controlled by the left hemisphere) pointed to a picture of a chicken. His left visual field, projecting to the nonverbal right hemisphere, was shown a snowy scene, and his left hand (controlled by the right hemisphere) pointed to a picture of a shovel. So far so good. But, when the individual was asked why he was pointing to the shovel, his response was, “You’re going to need a shovel to clean out the chicken coop.” From experiments like this one, Gazzaniga and colleagues came up with the explanation that our left hemisphere acts as an interpreter, explaining to others—and our conscious selves—why we do things (Gazzaniga, 2015).
It is not only individuals with split brains, however, who may not be aware of why they act in a certain way. In their classic paper, “Telling More Than We Can Know,” Nisbett and Wilson (1977) reviewed evidence that any of us may be unaware of stimuli that influence our responses. This idea was more recently articulated by Carruthers (2011), who argued that we do not have any more access to most of our own thoughts than we do to anyone else’s, and that we must use the same tools that we would use to discern how others think to discern our own thoughts.
These experiments by Gazzaniga (2015) and Nisbett and Wilson (1977) make it clear that what we consciously explain to others—and our selves—may not always be the true explanation. Similarly, confabulation by individuals with memory impairment, such as those with Korsakoff syndrome, may simply represent the conscious interpreter doing its best to make sense of the world with partial and outdated information.
To review, we believe that unconscious brain processes are ultimately responsible for our decisions and actions. If we consider our decisions and actions, most of the time we will have a ready explanation for them—and most of the time, our explanation will be correct. But sometimes, as these examples demonstrate, the explanation we tell our selves will be incorrect. This phenomenon is likely one reason why people use the psychological defense mechanism of rationalization—in this case, because they are not consciously aware of the truth, but only of the narrative that their conscious mind creates.
Willpower and Difficulty Controlling Behavior
Perhaps some of the most obvious examples that we do not have direct conscious control over our decisions and actions relates to willpower and the ability to control behavior. We have all had the occasional experience of finding it difficult to control our behavior. For example, we may have promised our selves that we are going to have just one spoonful of chocolate ice cream but, the next thing we know, the entire container is empty. While we were eating it, we may have said to our selves, “I shouldn’t be doing this. I should put the spoon down and put the container back in the freezer.” In this situation, it appeared that we were simply watching our selves eat the ice cream and that we had little control over our actions.
Our memory theory of consciousness helps us understand why it sometimes appears that we have little or no control over our own actions—because it is actually our unconscious brain processes that are in control. We now also understand why it appears that we are merely watching our selves perform actions that we do not consciously wish to do—because our conscious self is not participating directly; it watches memories of our actions (via our special video screen sunglasses as we are riding on our unconscious horse or via the Cartesian theater, depending on which metaphor is preferred). In fact, our memory theory of consciousness suggests that the sense that we are watching our selves act is a more accurate depiction of reality than the sense that we are consciously engaged in decision and action.
How the Conscious Mind Can Influence Decisions and Actions
Just because we argue that it is unconscious brain processes that ultimately decide and act does not mean that we believe that our conscious mind cannot influence those decisions and actions. Because conscious and unconscious processes are both occurring in the brain, it should not be surprising or mysterious that they interact, even if we do not understand the brain physiology underlying those processes and interactions. As Kahneman described so eloquently in his work with Tversky, we frequently have fast, unconscious System 1 “gut instinct” choices competing with slow, conscious System 2 logical choices (Kahneman, 2011). Luckily, for both the individual and the community, the conscious System 2 logical choices often win out for important choices, and the best decision is made. We would simply argue that, when that happens, the conscious System 2 “convinces” the unconscious System 1 to make the rational choice. The conscious memory system then remembers that the desired decision was made and thinks that it made the decision itself—even though it just remembered the decision that was actually made unconsciously. (The rider succeeds in cajoling the horse to go up the steeper path in order to avoid the dangers on what appears to be the easier route. The rider thinks that they made the decision, but where the rider goes is always up to the horse.)
We would like to make several points about free will. First, just because our decisions and actions are ultimately made unconsciously does not mean that we do not have free will—or, at least, not any more than if we made our decisions and actions consciously. The implications regarding determinism are no different whether decisions are initiated consciously or unconsciously. Second, as we just discussed, our conscious mind can cajole our unconscious self into making the best decisions in particular instances and can change the tendencies of our unconscious self over time. Third, if major life decisions are made slowly, over minutes, hours, days, or longer, these important decisions will almost certainly have input from both our conscious mind and our unconscious brain processes. Lastly, our memory theory of consciousness makes no predictions about whether our decisions and actions are determined on a microscopic scale, so on that point, we will not comment other than to say that we believe that we must act as if we have free will.
Who Are We?
One important issue that we have been skirting around is who, exactly, are we? Are we our conscious mind, our unconscious brain processes, both, or neither? The obvious answer is the correct one: We are both, of course. However, we would argue that the elements of our personality, our character—our innate tendencies of how we interact with the world—are determined by our unconscious brain processes, which we can call our unconscious self. Our conscious mind, on the other hand, is the part of us that does the conscious thinking, remembering, feeling, and calculating, and carries out other conscious processes. Note that our unconscious self certainly has different parts or aspects (Cisek, 2019). One part may crave chocolate ice cream while another part may wish to look svelte in the mirror.
Understandably, our conscious mind believes that we are primarily our conscious self, as our conscious mind spends most or all of each day alone with its self. From the perspective of everyone else interacting with us, however, we are primarily our unconscious self. Do not forget: We believe that most of our decisions and actions occur automatically without conscious input, and only a few have an element of conscious control. This means that, most of the time, the world is interacting with our unconscious self. Thus, we experience self and other as a mix of conscious and unconscious perceptions and actions. We almost certainly overestimate how much of the mix is conscious.
We Can Change Who We Are
Just because a large part of who we are is our unconscious self does not mean that we cannot change and improve our selves over time. As educators and clinicians, we most certainly believe that people can grow, learn, and change for the better. For example, perhaps we have noticed that we have a tendency toward selfishness. In contemplating this tendency along the lines we have discussed in this paper, we realize that it stems from one aspect of our unconscious self. Our conscious self, as well as other aspects of our unconscious self, may feel embarrassed about this tendency. To create a change, the parts of our unconscious self that desire a change can use conscious System 2 processes to create a conscious plan that may include both large actions in the world (such as sharing our time or money with charitable organizations) and small, everyday acts of selflessness (such as being considerate to those around us and helping strangers in need). If these actions are rewarded by positive feelings, our nonconscious memory systems (such as operant conditioning) will reinforce the actions such that they become part of our character—that is, our unconscious self.
This ability to change our character—the way we tend to interact with the world—is relevant to moral responsibility. If we find that our character tends toward bad behavior and we can change our character, then we are responsible for making ourselves into a better person. For instance, a parent may have an obligation to develop the (unconscious) tendency to engage in caring behaviors toward their children (cf. Björnsson and Brülde, 2017).
The memory theory of consciousness points us toward a mixed type of ethics that includes a healthy dose of virtue ethics. Most of our actions arise from our unconscious self, such that the proper object of moral evaluation is character, in line with the virtue ethics that are familiar from Aristotle. Other actions will be guided by conscious deliberation. Each of these deliberate actions will be subject to moral evaluation individually, which is consistent with deontological (“rule-based” or “duty-bound”) or consequentialist (“outcome-based”) ethics.
We Can Control Our Behaviors and Actions
In the same way that we can change who we are by consciously deciding to change our unconscious self, we firmly believe that we can control our actions. Perhaps this is obvious now that we have explained that we are both our conscious and our unconscious selves. But, even if we are just considering our conscious self, we believe that our conscious self can cajole and convince our unconscious self to make the decisions and take the actions we desire.
In fact, we hope that the viewpoint regarding consciousness expounded on in this paper will help everyone (including educators and clinicians) understand that in order to change our actions, we need to change both our conscious and our unconscious selves. Furthermore, to change our unconscious self may require the use of nonconscious training such as operant conditioning, skill learning, and the like. For example, to train our unconscious self to put our keys in the same place each day so we will not spend hours hunting around the house for them, we initially need to set up conscious reminders until procedural memory takes over and our unconscious is trained (ie, until it becomes a habit).
Returning to our one spoonful of chocolate ice cream example, for most people, it is not enough to simply consciously desire to stop after eating one spoonful using willpower. Instead, we will need to use our conscious mind to set up systems to either reduce the temptation of eating the entire container (perhaps dividing the container into one-spoonful amounts each on a plastic spoon, with one in front and the others in the back of the freezer) or reward our unconscious self for stopping after one spoonful with something greater than the reward that the chocolate ice cream itself would produce.
The way our memory theory of consciousness relates to behavior can enable us to describe heroic actions from a new vantage point. Some actions seem heroic to onlookers but come easily to the individual because of training or experience. For instance, running into a burning building to save a child is “just part of the job” for a firefighter. Other actions can be called heroic because they require overriding our System 1, unconscious character. We know that this type of action is possible. Indeed, the underdog characters in popular books and movies are often portrayed as overcoming certain unconscious or characterological tendencies (such as trepidation or cowardice) in just this way. Neville Longbottom from the Harry Potter books is an example.
Consistent With the Principles of Cognitive Behavioral Therapy
Everything we are suggesting in the ways in which our conscious mind can influence and train our unconscious self is consistent with the principles of cognitive behavioral therapy. Cognitive behavioral therapy explicitly recognizes that psychological problems are commonly based on both “unhelpful ways of [conscious] thinking” and “[unconscious] learned patterns of behavior” (American Psychological Association, 2017, text in brackets added). Moreover, cognitive behavioral therapy works both to change conscious thinking patterns and to use role play to change patterns of behavior. We believe that cognitive behavioral therapy is a successful therapy because it directly addresses both the conscious mind and the unconscious self, using techniques that each can understand.
Understanding Mindfulness Therapy
We can also understand mindfulness therapy in this context. By focusing attention on a process that usually occurs without our noticing, mindfulness can bring what is usually unconscious to the attention of the conscious self. That allows System 2 deliberate processing, deliberate evaluation, and deliberate changes in how we respond. To achieve the individual’s goals for the therapy, the new responses can remain in the domain of the conscious self or be adopted by the unconscious System 1 self.
Ethical Decisions and Actions
It follows from our discussion that an ethical decision or action may stem entirely from our unconscious self, or from both our conscious and unconscious selves. If we are in a horrible situation, we may jump in front of our child to save them from a bullet or other attack. Such actions would typically be automatic and too fast for conscious thought and deliberation. We make many small ethical decisions each week using our fast System 1 unconscious processing. At other times, we engage in slow, deliberate, System 2 conscious thinking to determine the appropriate ethical course of action; perhaps stepping down from a committee in protest if we strongly disagree with their decision, despite the loss of prestige and revenue that it would entail. Another example would be consciously choosing a less remunerative profession based on the amount of good it could produce in the world when certain aspects of our unconscious self would have chosen a more remunerative profession based on the amount of money or other comforts it could provide. Such consciously informed decisions would, of course, require our conscious mind to cajole and convince our unconscious self to make the actual decision.
Teaching Ethical Decisions and Actions
Our theory that all decisions and actions are ultimately made by the unconscious self has implications for the pedagogical methods that we use to teach ethics to our children and others in society. Teaching the theory behind ethics and ethical decisions and actions is helpful but not sufficient, as such theories will only inform the learners’ conscious minds and not their unconscious selves. Just as learning a new procedural skill first requires conscious memorization of a sequence and then practice of the actual motor/procedural skill so that the learning occurs in the unconscious self, so should ethical education involve not only conscious thinking and memorization, but also the visceral, emotional, empathetic learning that is acquired by doing. For example, children can participate in role-playing various actions and situations in order to feel the emotional consequences that different actions engender.
These ideas speak to the importance of ethical approaches involving fictional and autobiographical narrative insofar as these approaches support the integration of conscious rules with unconscious responses. These ideas also suggest that we may want to examine the ethics curricula in professional training (such as in medical, nursing, education, and law schools) so that we avoid a one-and-done, isolated ethics course in favor of an ethics-across-the-curriculum model.
Understanding Hurtful Actions
If, as we argue, decisions and actions are made unconsciously, then we must work together as a society to use this understanding to reduce the occurrence of hurtful acts through pedagogical methods, such as role-playing. We should not excuse those people who commit hurtful acts, but we should also have greater sympathy for them because we understand that they may be acting on unconscious impulses. Nonetheless, in order to live together in society, individuals need to learn to control their unconscious selves and thus their behavior through cognitive behavioral therapy and other ethical methods. (Although the horse determines where it will go, the rider needs to learn to tame and control the horse.)
Developing Conscious Control of the Unconscious Mind: The Importance of Mindfulness
Mindfulness has recently been touted as the key to everything in our society from avoiding burnout to achieving enlightenment. Initially, we raised the issue of why mindfulness is hard, and then we provided our explanation that mindfulness is hard because our conscious mind did not develop to be in control of our thoughts; rather, the unconscious parallel-processing brain developed our conscious mind simply as a necessary part of our conscious memory system. That is, we argue that our conscious mind developed as a tool to be used by our unconscious brain to help it predict the future and make decisions accordingly. It is for this reason that we believe that it is difficult for us to consciously control our conscious thoughts—because normally, our unconscious brain is in control of our conscious thoughts!
If, however, we wish to achieve more control over our decisions and actions, it is helpful to have more control over our conscious thoughts. Once we have control over our conscious thoughts, we can use our thoughts to cajole and convince our unconscious self to make the decisions we consciously desire, whether that is eating only one spoonful of chocolate ice cream, leaving work problems at work when we come home to our family, or stopping the flow of thoughts that run through our mind when we are trying to sleep. Because mindfulness is one method of improving our ability to consciously control our thoughts, we believe that practicing mindfulness can help us control our decisions, behaviors, and actions as well.
Unlock the Power of the Unconscious Self
Sometimes we have the experience that someone asks us a question and, without thinking, we start to answer, only to be surprised at the words that leave our mouths. In this circumstance, we have stepped back and reflected, “Oh, I guess that is what I think about that. …” We believe that this situation is an example of how we may learn what our unconscious self thinks about an issue, and it is just one way that we can use our unconscious brain processes to our advantage.
Put the Unconscious to Work
Most people have had the experience of trying to come up with a name or an answer to a question and, after the conversation has finished, the name or answer then pops into consciousness. We can actually task our unconscious to work on a problem, and our unconscious brain may (or may not) help us solve it. By simply saying to our selves, “OK, I need to recall that piece of information,” and then turning to other matters, we may find the information coming to us spontaneously later that day, often within minutes. Another method we use is to review some of the work that we need to do the following day just before sleep. For example, before sleep, we will generally review the last section of writing we have done as well as our notes or our plan of what we will be writing the following day. Based on the work of Sanders, Beeman, Paller, and others (Paller et al, 2021; Sanders et al, 2019), we assume that this little bedtime preparation facilitates our unconscious brain processes while we are sleeping. In any event, when we sit down at our computer the next morning, the writing proceeds easily.
Bring Unconscious Processes to Conscious Awareness
Sometimes we are in a meeting and notice that a person appears scared, angry, sad, excited, or happy. We are all fairly good at detecting basic emotions in ourselves and others using either unconscious or conscious processes (Calvi et al, 2020; Ochsner et al, 2004; Smith and Lane, 2015). What about more complex states such as bluffing, lying, manipulating, seducing, protecting, defending, and supporting? These can be a bit more difficult to discern consciously, but we are aware of many of them some of the time at an unconscious level. We can train ourselves to bring these states that other people are exhibiting to our consciousness by consciously analyzing the individual’s tone of voice, body language, eye contact, and others. We can also tap into our unconscious perception by noticing how the person makes us feel. Do we feel sorry for them? Do we like or dislike them? Are we afraid of them? Do they make us angry? Do they make us smile—and if so, why? By analyzing our reactions to people and events, we can bring at least some of our unconscious perceptions to consciousness.
Stereotype Threat and Other Unconscious Biases
One of the most important reasons to bring unconscious thoughts and emotions to conscious awareness is to reduce the effects of stereotype threat and other unconscious biases that we are all susceptible to. Stereotypes and biases have been shown to impact the careers of women and minorities (Grewal et al, 2013), in addition to having a detrimental effect on many other aspects of society. Education and training have been shown to increase awareness of these issues and provide concrete steps that individuals and institutions can follow to reduce the harmful effects of stereotypes and unconscious biases (Grewal et al, 2013; Rodriguez et al, 2021).
We sometimes have the experience that it is hard to get a word in edgewise in a conversation, as if we are half a second behind, while at other times it is easy and effortless to contribute to the discussion. We believe that, at least in part, it depends on whether we are responding consciously, with the built-in perceptual delay of approximately half a second, or unconsciously, which has no delay. If this speculation is correct, the more consciously determined we are to make a specific point in the conversation, the more likely we are to have trouble doing it because we will be experiencing the perceptual delay. We will be more able to break into the conversation if we relax and let our reactions and responses be more spontaneous.
Can our language be controlled by unconscious brain processes? Or do speaking and writing always involve not only conscious awareness but conscious participation in such activities? We hinted at aspects of these questions previously, discussing how we can sometimes be surprised by the words that come out of our mouths. This example alone is perhaps sufficient to make the claim that unconscious brain processes can control our language—how else could our conscious mind be surprised at what we just said?
Another example relates to when we are giving a lecture that we have given many times before. Here, we have noticed that two related and interesting phenomena sometimes occur. The first is that we often enter a mode where we, as our conscious selves, can watch and listen to our selves talk on what appears to be automatic pilot—we at least do not feel as if we need to use any conscious effort to make the words come out of our mouths. It is as if we are not only in the Cartesian theater watching the world go by, but we are also able to see our selves on the stage! The second is that, during such a time when we can step back and watch our selves lecture, we sometimes get distracted, start thinking of something completely different, and do not pay attention to our selves lecturing. When our conscious awareness returns to our lecture, we find that we cannot remember what point we have just made, and whether we should be making the point that is on the slide on the screen or have already done that and should move on to the next slide.
Thus, speaking without conscious control or even awareness is not only possible, but probably common. As with other perceptions, decisions, and actions, if we are not consciously aware of our speaking, we will not be able to remember what we have just said.
In the Zone: Playing Sports and Musical Instruments
As discussed earlier, reaction times are often too fast in sports and music to be consciously controlled. Thus, many athletes and musicians perform using their unconscious System 1 brain processes. In fact, many athletes and musicians perform much better when they use only their unconscious processes to control their behavior. This unconscious performance is often referred to as being “in the zone,” which the Oxford English Dictionary defines as “a state of perfect concentration leading to optimum mental or physical performance.” We suggest that being “in the zone” means that unconscious brain processes are given full rein to make the right, often precise or creative, choices much faster and more accurately than would be possible with conscious control. This level of unconscious expertise is, of course, only possible when these unconscious brain processes have been trained through thousands of hours of practice. But, once that training has taken place, the unconscious brain processes can be let loose! Conscious control is as likely to interfere with performance as it is to help.
We are not saying that conscious processes have no role in sports or music. Some of the best athletes, such as the tennis champions Serena Williams and Roger Federer, are known to consciously modify their strategy when their current approach is not producing success. Similarly, some of the most creative music, such as that from Miles Davis and Lady Gaga, appears to result from the interaction of conscious and unconscious brain processes working together.
Art and Insight
Many—perhaps most—creative endeavors, from painters to novelists, likely result from the combination of conscious and unconscious brain processes working together. Additionally, the appreciation of beauty and other aesthetics in any sensory modality (vision, hearing, smell, taste, touch) may be driven primarily by unconscious properties, which may be why aesthetic appreciation is so difficult to describe using words. We suspect that creative insights in other fields, whether psychology, biology, chemistry, physics, mathematics, and others, often arise from unconscious processes in a brain that has been prepared by consciously struggling with the same or related problems for some time. As Louis Pasteur famously said, “Dans les champs de l’observation le hasard ne favorise que les esprits prepares,” which is often colloquially translated as “Chance favors the prepared mind.” Such scientific insights that likely arose through the combination of conscious and unconscious processes include those of Alice Ball, Rachel Carson, Marie Curie, Marie Daly, Jennifer Doudna, Albert Einstein, Jane Goodall, August Kekulé, Ada Lovelace, Isaac Newton, and many others. Some of the insights of these people might be because they are somehow able to see the world more the way it actually is, rather than the way the conscious mind suggests that it is. In other words, some of these individuals may have had more access to their unconscious brain processes.
Hume, for example, asserted that the self is “nothing but a bundle or collection of different perceptions” (Hume, 1978, p. 252), and that we can have no idea of any subject to which these perceptions belong. It is simply that we “bundle” these experiences together that they appear continuous. It is possible that some of Hume’s (1978) insights may be related to his glimpsing some of his unconscious brain processes.
Here, we outline some possible methods that can be used to test different aspects of our memory theory of consciousness. We will first discuss methods that may be able to prove or disprove the theory that consciousness developed as a memory system, and then methods that examine the idea that each region of the cerebral cortex is sufficient to make one aspect of consciousness possible.
Methods to Evaluate Consciousness as a Memory System
We have argued that consciousness developed as part of a memory system and that all explicit forms of memory (sensory, working, episodic, and semantic) are part of that unified system. One line of investigation would therefore be to examine whether consciousness is truly bound up with, and an integral part of, these forms of memory. Are there situations in which explicit memory can exist in the absence of conscious awareness? Conversely, can there be any type of conscious awareness without involvement of one of these forms of explicit memory? Are there correlations (or not) between explicit memory performance and conscious perception abilities? Are there correlations between subjective measures of explicit memory and subjective measures of consciousness? Such experiments would provide important insights into this issue.
Experiments that are similar to those performed by Schneider and colleagues (2021), which attempted to show engagement of the episodic memory system with and without conscious awareness by using different types of visual masking, may be one fruitful line of research. Other paradigms that contrast conscious versus unconscious processing of stimuli may also be useful (for reviews, see Breitmeyer, 2015, and Timmermans and Cleeremans, 2015).
The key consideration in all of these paradigms would be the relationship between conscious awareness of, and explicit memory for, the stimulus. These include paradigms that use change blindness (eg, Simons, 2010), attentional blink (eg, Sy et al, 2021), visual crowding, continuous flash suppression, perceptual fading, motion-induced blindness, reversible figures, binocular rivalry, perceptual filling in (eg, Davidson et al, 2020), and visual perception of degraded objects (eg, Levinson et al, 2021). False memory paradigms may also be useful (Nichols and Loftus, 2019; Schacter and Dodson, 2001); these experiments often produce a vivid, confident, conscious recollection of an item or experience that never occurred (Sikora-Wachowicz et al, 2019). Visual imagery experiments, in which a conscious image is produced in the absence of a percept, may also be useful (eg, Moulton and Kosslyn, 2009). Lastly, experimental paradigms involving sleep and memory may be particularly useful (eg, Paller et al, 2021), as dreams appear to be one area in which consciousness and memory may not be co-occurring.
Subjective measures have been used to fractionate memory into different components, such as the remember–know (Tulving, 1999) and recollection–familiarity distinction (Yonelinas, 1994). Subjective measures ask participants to use introspection (as in the case of remember–know) or to provide confidence judgments. Confidence judgments have been successfully used to create receiver operating characteristic curves to estimate the degree to which participants use recollection and familiarity during a task (Yonelinas, 1994). Process-dissociation procedures can also be used to separate components of memory (eg, automatic vs intentional use [Jacoby, 1991]).
Subjective measures have also been used to evaluate conscious awareness (for review, see Timmermans and Cleeremans, 2015). These measures include the perceptual awareness scale (Ramsøy and Overgaard, 2004), in which participants rate stimuli as glimpsed, almost clear, or clear; the continuous visual analog scale (Sergent and Dehaene, 2004), in which participants rate stimuli on a continuum from not seen to maximally visible; the rule awareness scale (Wierzchoń et al, 2012), in which participants rate how aware they are of a rule; postdecision wagering (Persaud et al, 2007), whereby participants wager money on their response; confidence ratings (Dienes et al, 1995), whereby participants rate their conscious awareness; and a feeling of warmth (Metcalfe, 1986), which is similar to confidence ratings but lends itself to a more intuitive response.
Methods to Evaluate Order and Timing Issues Around Consciousness
Regarding issues of the order and timing of consciousness, investigations of chronostasis (eg, Melcher et al, 2020) and postdictive effects (for reviews, see Herzog et al, 2020; Michel and Doerig, 2021; and Sergent, 2018) would likely be informative. Such paradigms include the cutaneous rabbit illusion (Geldard and Sherrick, 1972), illusory and invisible audiovisual rabbits (Stiles et al, 2018), color fusion effects (Pilz et al, 2013), the color phi illusion (Keuninckx and Cleeremans, 2021), the sequential metacontrast paradigm (Otto et al, 2006), rapid serial visual presentation (Akyürek and Wolff, 2016), and cues that are presented after a faint Gabor grating (Sergent et al, 2013). Key in these paradigms would be the relationship between explicit memory and the conscious postdictive experience.
Disrupting Consciousness and Memory
Another line of research that may be fruitful would be to disrupt either explicit memory or conscious awareness and evaluate whether the other is also disrupted. For example, will medications that are known to impair explicit memory (eg, anticholinergics, benzodiazepines) impair consciousness as well? Will medications that are known to impair consciousness (eg, sedatives, anesthetics) impair memory as well? Will a TMS pulse that disrupts explicit memory also disrupt at least one aspect of consciousness and vice versa?
Neural Signals of Consciousness and Memory
Although currently there is no consensus as to which physiological markers correlate with conscious experience, one can still evaluate whether such markers correlate with both one domain of conscious awareness and explicit memory (rather than correlating with one but not the other). Putative markers of consciousness—such as EEG activity and its derived components, ERPs—may be of particular interest.
Methods to Evaluate the Theory That the Cerebral Cortex Is Where Consciousness Occurs
We have argued that not only do all areas of the cerebral cortex contribute to consciousness, but that each region of the cortex may be autonomously conscious of its own particular aspect of consciousness. This theory is contrary to those of a number of other researchers, some of whom have suggested that the frontal lobes are the minimally sufficient regions necessary for consciousness (Brown et al, 2019; Murray et al, 2020) and others who have suggested that the parietal and occipital cortex are the regions that are minimally sufficient for consciousness (Lamme, 2015; Lee et al, 2019; Tononi et al, 2016).
Methods for evaluating the anatomical aspect of our theory of consciousness and that of other theories can include studies of individuals with brain lesions or dementia, studies of simulated lesions using TMS, brain imaging studies that provide anatomical localization using fMRI or PET, brain imaging studies that provide information about the timing and physiology of brain processes using EEG and ERPs, derived EEG measures such as the bispectral index (BIS) that is used by anesthesiologists, and other derived measures such as the PCI.
Different Domains of Consciousness May Have Different Neural Correlates
Because we believe that different regions of the cortex can be autonomously conscious and that the consciousness awareness (the conscious phenomena or qualia) provided by each region will be different, there is no reason that the physiological markers of these different brain regions need to be the same. Therefore, from our perspective, the task is to look for physiologic markers of conscious versus unconscious processes for each cognitive domain (eg, conscious visual awareness vs unconscious visual sensation) rather than conscious versus unconscious processes of the brain as a whole. Even within a cognitive domain, different experimental paradigms may lead to different conclusions regarding the neural correlates of consciousness due to the specifics of the paradigm and how consciousness is measured and reported (eg, Cohen et al, 2020).
If this idea is correct, one fruitful project would be to try to separate different domains of conscious awareness (eg, visual, auditory, olfactory, gustatory, tactile, etc.) and perhaps also differences within a domain (eg, visual perceptual fading, visual perceptual filling in, and visual perception of degraded objects, etc.) and then try to determine the neural correlate of consciousness for each. Some scientists and philosophers would argue that amodal thoughts (ie, abstract thoughts completely separate from sensation) can also be conscious and should be added to this list, whereas other scientists and philosophers would argue that that is not possible (Carruthers, 2015).
Fixed and Temporary Brain Lesions
One straightforward way to evaluate whether our cortical hypothesis of consciousness is correct is to study individuals with brain lesions resulting from strokes or surgical resections. The advantages of studying such individuals include that their lesions are generally constant over time and, using structural MRI combined with measures of white matter integrity (such as diffusion tensor imaging), the anatomical extent of the lesions and their connections can be readily ascertained. Studies of individuals who happen to have symmetric bilateral lesions may be particularly informative. Such individuals can then be evaluated for different domains of consciousness via structured interviews, questionnaires, and experimental paradigms.
A similar approach can use TMS to produce transient brain lesions in various cortical areas. Advantages include that lesions can be produced on demand and, when combined with structural MRI for localization, in the precise area that is wished. Lesions can also be bilateral. Disadvantages include that these lesions are transient and that deep brain structures, such as the thalamus, hippocampus, and insular cortex, are difficult to stimulate.
Another promising approach to ascertaining the anatomical regions associated with certain aspects of consciousness is to combine these methods by evaluating an individual with a known, unilateral, fixed brain lesion and then using TMS to temporarily inactivate the reciprocal cortex on the opposite hemisphere. In all of these situations, one of the key questions is whether there are any bilateral cortical lesions that produce complete unconsciousness. Our theory would predict that bilateral lesions would impair specific domains of consciousness, but not other domains of consciousness. The hypotheses of several other researchers (eg, Boly et al, 2017; Michel and Morales, 2020; Odegaard et al, 2017) predict that all domains of consciousness will be impaired when certain bilateral cortical regions are not functioning.
Dementia as Bilateral and Multifocal Brain Lesions
Although one does not frequently think of individuals with dementia as helpful for addressing issues of focal brain dysfunction, this approach has several advantages. First, many dementias affect brain regions bilaterally. Although the impact of the disease is not generally identical between the two hemispheres, it is often roughly equivalent. For example, individuals with AD may show roughly symmetric bilateral atrophy of the hippocampi, anterior temporal lobes, and parietal lobes. Individuals with posterior cortical atrophy may show roughly symmetric atrophy of the parietal and occipital lobes. Some individuals with behavioral variant frontotemporal dementia show roughly symmetric atrophy of the frontal lobes.
But, the real opportunity of studying individuals with cortical dementias comes when the precise degree of atrophy is measured by cortical thickness, for example, using structural MRI. When the degree of atrophy is measured along with consciousness-related variables including structured interviews, questionnaires, experimental paradigms, and potentially EEG, ERP, BIS, PCI, and other measures, which cortical areas are important for specific domains of consciousness can be ascertained. This approach has been used successfully in parcellating different aspects of memory (Wolk et al, 2011).
fMRI and PET
fMRI and FDG (fluorodeoxyglucose) PET studies are the standard tools that are used to evaluate the function of various brain regions. Such tools have been used to evaluate consciousness and can yield valuable insights when they are combined with appropriate experimental paradigms. Paradigms that contrast conscious versus unconscious perception are key (eg, Levinson et al, 2021; for reviews see Mashour et al, 2020, and Michel and Morales, 2020). Similar paradigms can be used to evaluate other domains of consciousness by contrasting conscious versus unconscious perception in auditory, olfactory, tactile, and perhaps even gustatory modalities. New paradigms can be developed to evaluate conscious versus unconscious perceptions, decisions, and actions. Taken together, such experiments will enable our hypothesis that all cortical regions participate in their own domain of consciousness to be tested.
EEG and Derived EEG Measures Such as the BIS, ERPs, and PCI
Although a scalp EEG does not provide good spatial localization, it has the advantage of measuring the electrical activity of the cerebral cortex. Raw EEG data, such as the proportion of different EEG rhythms, is already used to determine various levels of consciousness, related both to sleep and to coma stages. Derived EEG measures, such as the BIS, are used by anesthesiologists to measure the depth of anesthesia and reduce intraoperative awareness. Both raw and derived EEG measures can therefore provide a measurement of consciousness that may be informative regarding our and others’ hypotheses about various aspects of consciousness.
ERPs are averages of the EEG signal time locked to specific auditory or visual stimuli. ERPs can provide information regarding the precise timing of events. ERPs have also been used to provide information regarding the cortical activity of a variety of different conscious and unconscious brain processes, including those related to attention, perception, working memory, episodic memory, and language. We believe that important insights can be gained when ERPs are measured in experimental paradigms that evaluate conscious versus unconscious brain processes.
Global neuronal workspace theory hypothesizes that consciousness can be measured by a late positive ERP component that occurs ∼300 ms after a stimulus, which is often called a P300 or P3 (Mashour et al, 2020). This component is thought to be the same as the old–new parietal effect that in memory research is often termed the late positive component. Using this ERP component to detect conscious awareness is consistent with studies of conscious awareness of remembering (Ally et al, 2008), and with the ∼300-ms delay suggested by our memory theory of consciousness. Other researchers, however, have suggested that the P3b tracks what observers are reporting—not what they are perceiving (Cohen et al, 2020). Regardless of the debate regarding this particular ERP component, we believe that ERPs will be helpful in the search for neural correlates of consciousness. We believe it is likely that different ERP components will reflect different domains of consciousness—one for the conscious perception and one for the action of reporting.
PCI can be thought of as a special type of ERP, one that is triggered by TMS pulses rather than by auditory or visual stimuli. Additionally, rather than simply averaging the resultant EEG signal to produce the ERP, the EEG data are evaluated regarding its integration (a measure of how wide the perturbation of the TMS pulses spread) and its information (a measure of how compressed its spatiotemporal pattern of activity can be). A single number is produced. PCI has been shown to be able to reliably differentiate individuals who are awake from those whose consciousness is diminished due to sleep, anesthesia, or coma (Casali et al, 2013).
Evaluating MRI, FDG PET, EEG, BIS, ERP, and PCI in Individuals who May Exhibit Impaired Consciousness
Lastly, if we are correct that there are individuals who show impairments of consciousness (as described in Disorders of Consciousness), then the prediction naturally follows that measuring aspects of their consciousness using neuroimaging, neurophysiological, and computational derived techniques should yield differences between these individuals and those with normal consciousness.
One approach to answering the question of how large a cortical region would need to be in order to support independent consciousness would be to use structural MRI techniques to measure cortical areas in individuals who demonstrate phenomenologically impaired consciousness of a particular modality (such as visual consciousness) due to a brain lesion, degenerative disease, or other pathology. The relevant structural maps and total cortical volumes in these individuals could be compared with those of other individuals who also experienced damage to the relevant brain regions but have intact consciousness in that modality.
CONCLUSIONS AND LIMITATIONS
Our central claim is that consciousness is essentially and originally part of explicit memory. We experience the world progressing serially because our conscious memory system creates a linear, coherent stream of experiences from our unconscious, parallel brain processes. We believe that our memory theory of consciousness is useful (and perhaps correct) because it helps explain phenomena that have been recognized as long-standing puzzles for previous theories, such as postdictive effects. We have shown how our memory theory of consciousness helps us understand clinical syndromes, experimental studies, and everyday experiences. We have also hypothesized that regions of the entire cerebral cortex are the functional units that make consciousness possible.
We are hopeful that this paper provides the framework for a fruitful line of theoretical, observational, and experimental work that can prove or disprove each of the hypotheses that we have put forth. As is implicit in that statement, we are well aware that many—perhaps even most—of the hypotheses that we are proposing may turn out to be incorrect. What we are confident of, however, is that research using the methodologies that we and other researchers have outlined to test our hypotheses will move forward the cognitive neuropsychology, experimental psychology, and cognitive neuroscience of consciousness studies, bringing us closer to understanding the fundamental nature and anatomical basis of consciousness.
In addition to the possibility that some of our hypotheses are wrong, we must also acknowledge that our hypotheses only discussed several small aspects of consciousness and ignored many of the most important parts of any complete theory of consciousness, such as the so-called hard problem (how a collection of biological material can produce the subjective experience that we call consciousness). Nonetheless, by careful observation and well-designed experiments, examining this memory theory of consciousness may move us toward a time in the future when such questions will seem quaint, similar to questions about what constitutes the life force that living beings have or how light travels through the ether.