Share this article on:

Chronic pain impairs cognitive flexibility and engages novel learning strategies in rats

Cowen, Stephen L.a,*; Phelps, Caroline E.b; Navratilova, Editab; McKinzie, David L.c; Okun, Alecb,c; Husain, Omarc; Gleason, Scott D.c; Witkin, Jeffrey M.c; Porreca, Frankb

doi: 10.1097/j.pain.0000000000001226
Research Paper
Editor's Choice

Cognitive flexibility, the ability to adapt behavior to changing outcomes, is critical to survival. The prefrontal cortex is a key site of cognitive control, and chronic pain is known to lead to significant morphological changes to this brain region. Nevertheless, the effects of chronic pain on cognitive flexibility and learning remain uncertain. We used an instrumental paradigm to assess adaptive learning in an experimental model of chronic pain induced by tight ligation of the spinal nerves L5/6 (spinal nerve ligation model). Naive, sham-operated, and spinal nerve ligation (SNL) rats were trained to perform fixed-ratio, variable-ratio, and contingency-shift behaviors for food reward. Although all groups learned an initial lever-reward contingency, learning was slower in SNL animals in a subsequent choice task that reversed reinforcement contingencies. Temporal analysis of lever-press responses across sessions indicated no apparent deficits in memory consolidation or retrieval. However, analysis of learning within sessions revealed that the lever presses of SNL animals occurred in bursts, followed by delays. Unexpectedly, the degree of bursting correlated positively with learning. Under a variable-ratio probabilistic task, SNL rats chose a less profitable behavioral strategy compared with naive and sham-operated animals. After extinction of behavior for learned preferences, SNL animals reverted to their initially preferred (ie, less profitable) behavioral choice. Our data suggest that in the face of uncertainty, chronic pain drives a preference for familiar associations, consistent with reduced cognitive flexibility. The observed burst-like responding may represent a novel learning strategy in animals with chronic pain.

Rats with spinal nerve ligation were impaired relative to controls in flexibly adapting to changing environmental demands.

Departments of aPsychology and

bPharmacology, University of Arizona, Tucson, AZ, USA

cNeuroscience Discovery Research, Lilly Research Laboratories, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN, USA

Corresponding author. Address: The University of Arizona Life Sciences North, Rm 347 1501 N. Campbell Ave., Tucson, AZ 85721, USA. Tel.: 520 626 2615. E-mail address: (S.L. Cowen).

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

Received November 29, 2017

Received in revised form February 15, 2018

Accepted March 06, 2018

Back to Top | Article Outline

1. Introduction

Chronic pain induces morphological changes in the prefrontal cortex (PFC),5,18 a brain region associated with high-order executive functions such as attention, inhibitory control, and cognitive flexibility.3,32,70 Greater PFC volume and thickness correlate with improved executive function, and patients with frontal lesions exhibit impairment.3,70 Imaging studies have demonstrated reductions in grey matter in the PFC in patients with chronic pain.5,18 However, the possible consequences of pain-related morphological changes on behavior are unknown. One potential consequence is impaired executive function such as reduced cognitive flexibility.

Cognitive flexibility is the ability to adapt behavior to changing outcomes12 and is associated with increased resilience to negative life events,19 improved academic performance in children,17 as well as better overall quality of life.13 Patients with chronic pain may be impaired in cognitive flexibility, reflected by reduced performance on tasks such as the trail making test (TMT)30,57,67 and the Wisconsin Card Sorting Task (WCST).28,48,65,68 These tasks challenge subjects to adapt to changing rules for trial completion (TMT) or reward (WCST). Impaired WCST performance may be associated with reduced PFC function,6,44 although not exclusively.20,21,49,50 The impact of PFC on TMT is less clear.70 Assessing the impact of pain on behavior arises from the possible relationship of deficits to factors such as medications used for treatment, lifestyle, socioeconomic status, and other factors rather than to pain itself. In addition, the underlying neurobiology of pain-associated morphological changes in the PFC remains unclear, so that establishing causality of pain with cognitive function is difficult. For example, it is currently unknown whether subjects with chronic pain use alternate learning strategies that allow them to overcome cognitive deficits produced by potentially disrupted frontal function.

The use of preclinical models allows assessment of the influence of chronic pain on cognitive function under controlled conditions in homogenous subjects. Despite the reduced complexity of the rodent PFC, this region shares many homologous roles in executive control with primates (as reviewed in Ref. 32), and the morphology of the rat PFC has been shown to be similarly altered in chronic pain states.2,11,14,31,42,62 Rodents in chronic pain are unable to learn platform relocation in the Morris water maze36,46 or a bowl-digging attentional set-shifting task,37,38 suggesting impaired cognitive flexibility. However, the influence of other aspects of cognition and alternate learning strategies has not been studied in these tasks.

In the present study, naive, sham-operated, or nerve-injured rats were tested in an operant protocol of fixed ratio (initial learning), progressive ratio (motivation), reversal (adaptive learning), and variable-ratio reversal (cognitive flexibility and decision making). This design allowed measurement of multiple aspects of cognition, as well as testing for effects of chronic pain on coping strategies, memory consolidation, and motor impairment. Our data suggest that chronic pain promotes novel adaptive learning strategies that are not fully capable of ameliorating impaired cognitive flexibility.

Back to Top | Article Outline

2. Methods

2.1. Animals

Male, Sprague–Dawley rats (Harlan Laboratories Inc, Indianapolis, IN) weighing 250 to 350 g were housed in climate-controlled rooms on a 12-hour light–dark cycle. All animal procedures were approved by the Institutional Animal Care and Use Committee at Eli Lilly, and these procedures were in accordance with the guidelines set by the National Institutes of Health and the International Association for the Study of Pain. Animals were monitored throughout the duration of the study to reduce any unnecessary stress or pain. Subjects were divided experimentally into spinal nerve ligation (SNL) (N = 16), sham (N = 15), and naive (N = 12) populations. Animals were housed individually for the duration of the experiment (∼11 weeks) to control feeding conditions during training periods involving food restriction. During food restriction (∼4 weeks), animals were monitored, handled, and tested in the operant chambers by experienced personnel. Data on motivational responses were collected from the same cohort of rats and previously reported.51

Back to Top | Article Outline

2.2. Spinal nerve ligation

The L5/L6 surgical procedure was used to produce experimental neuropathic pain as previously described.33 The L5 and L6 spinal nerves (left side) were tightly ligated with 4-0 silk suture and the wound was closed. Sham-operated rats were prepared in the same manner as the SNL rats but without ligation of the L5/L6 spinal nerves. Naive rats had no surgery or other conditions placed on them. Animals were acclimated in the housing facility for a minimum of 7 days before experimentation. All rats were monitored for any signs of motor impairments and assessed for overall health. Spinal nerve ligation–induced allodynia was confirmed in all animals using von Frey filaments at postsurgery days 14 to 120.33 Compared with sham controls, SNL rats exhibited significantly decreased tactile withdrawal thresholds throughout the course of the experiment, confirming the presence of mechanical allodynia (tested on days 14, 21, 28, 35, 42, 56, 90, and 120 post-SNL). Mean withdrawal thresholds ranged between 3.9 to 5.2 g for SNL rats and 14.1 to 15.0 g for sham controls; F(1,208) = 988.3, P < 0.0001 for main effect of treatment; (see Fig. 3E in Ref. 50). Using a 6-g exclusion threshold for the presence of allodynia, no SNL rats were excluded from the study.

Back to Top | Article Outline

2.3. Operant conditioning

Each rat was trained on a series of operant tasks that assessed associative learning. Before training and on day 16 postsurgery, food restriction was implemented so as to reduce body weights to 85% to 90% of their free-feeding weights. Training began on postsurgery day 21 and the training regimen is summarized in Figure 1 and described below. The choice task (task 3) and the variable reinforcement task (VR, task 6) are the main targets of analysis; performance on the progressive ratio task (PR, task 2) has been reported previously.51

Figure 1

Figure 1

Back to Top | Article Outline

2.3.1. Lever-press acquisition (task 1) and progressive ratio testing (task 2)

Animals were first trained on a fixed-ratio (FR) task for 4 days (task 1), followed by training on a PR task (task 2). Training on these tasks began on postoperative days 21 to 23 when rats were placed in 2-lever operant responding chambers (12.0″ L × 9.5″ W × 8.25″ H, Med Associates, Fairfax, VT) and underwent training for 90 min/d on sessions 1 to 3 (Fig. 1A). During all training sessions, presses on the right lever were followed by a 2-second tone and delivery of 1 sucrose pellet (Dustless Precision Pellets [45 mg; Bioserv, Flemington, NJ]) for each press (FR1 schedule). Presses on the left did not result in food delivery during tasks 1 and 2. The maximal number of pellets delivered in a single session was 50, but not all rats reached 50 pellets on a given day. On the fourth training day, (postoperative day 24), the rats were again allowed to lever-press for food; however, the session length was decreased to 30 minutes. Rats that did not acquire 50 pellets on this day were excluded from the experiment (n = 2 rats). On day 25, the progressive ratio testing started with the reinforcement schedule set such that the number of lever presses required to receive a single pellet increased after each pellet delivery (PR schedule). The initial PR schedule (PR1) was as follows: 1, 1, 1, 2, 2, 2, 4, 6, 8, 10, 12, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, and 80. At the highest ratio, rats had to press the lever 80 times to obtain food. After completion of the PR1 schedule on day 26, food restriction in the home cage was stopped and rats had free access to chow until the end of postsurgery day 41. Rats were tested again on day 27 with the PR1 schedule. On day 28, the reinforcement schedule was made more challenging by increasing the ratios more rapidly. The progressive ratio schedule 2 (PR2) for day 28 was as follows: 1, 2, 4, 8, 16, 32, and 64.

Back to Top | Article Outline

2.3.2. Choice task (task 3)

Choice testing began on postsurgery day 29, which was the first day after the PR experiment (Fig. 1A). In this FR task, rats received a single sucrose pellet per lever-press of the right lever and 4 pellets per press of the left lever. Performance on this task was quantified as the percent of trials in which the rats selected the high-reward lever. Rats were tested without food restriction for the first 8 experimental sessions of task 3. After the eighth session, rats were food-restricted and tested for an additional 5 sessions. Experimental sessions lasted 60 minutes or 50 food pellet deliveries, whichever occurred first, after which the chamber became inactive and subsequent responding was without consequence. Food restriction during days 21 to 28 was used to speed training. Restriction was not present during the initial phase of task 3 (days 29-41) to eliminate potential antiallodynic effects of food restriction on SNL animals, as well as to improve translation between rodent and human studies because human patients with chronic pain are not food-restricted.

Back to Top | Article Outline

2.3.3. FR1 and cue association tasks (tasks 4 and 5)

By design, training on the choice task resulted in all animals developing a preference for the high-reward (left) lever (Fig. 2). To reduce the preference for the higher reward lever after task 3, the rats received 2 days of training where both levers were active and reinforced on an FR1 = 1 pellet schedule (task 4). Testing occurred for 30 minutes or until the rats received 50 total pellets. Rats were food-restricted for both experimental sessions. To further reduce bias for a single lever, these 2 FR1 sessions were followed by a light-following task (task 5) in which a light appeared over 1 randomly chosen lever. When the light was on, the lever was active and would deliver 1 pellet when pressed. After the lighted lever was pressed, the light randomly appeared over 1 of the levers again to signal a new active lever. The session lasted 30 minutes or ended at 50 food pellet deliveries, whichever came first.

Figure 2

Figure 2

Back to Top | Article Outline

2.3.4. Variable-ratio choice testing (task 6)

The variable-ratio (VR) task began on postsurgery day 56. In this task, both levers were active, and the right lever delivered 1 pellet per press on 8 of 10 presses, whereas the left lever delivered 4 pellets of reward on 3 of 10 presses. Consequently, the left lever was designated as the high-reward lever because the average reward per 10 presses was 12 pellets, whereas the average reward delivered on the right lever was only 8 pellets. Given the schedule just described, the variability in the VR schedule was low. This low-variance VR schedule was chosen to ensure consistent reinforcement probabilities. Experimental sessions lasted 60 minutes or 50 food pellet deliveries, whichever occurred first.

Back to Top | Article Outline

2.4. Statistical methods

All statistical analyses were conducted with Matlab 2017a. Differences in response rates between groups were assessed using the Student t tests or analysis of variance (ANOVA) with Tukey–Kramer or Bonferroni–Holm multiple comparisons corrections. Significance was set at α = 0.05 for all tests. Learning rates for the choice and probabilistic tasks (percent high-reward) were determined for each animal by computing the slope of the least-squares regression line that fit the session-by-performance data (Fig. 2A).

Within-session learning was quantified in 2 ways. The first approach measured learning as the difference between the percent in which the animals selected the high-reward lever during the past 1/3 of trials minus the percent of high-reward lever presses during the first 1/3 of trials (Percent High Reward Last Third − Percent High Reward First Third). The second approach involved fitting a least-squares regression line to the pattern of left–right responses per session coded as 0s and 1s (eg, 0 0 0 1 0 1…). The slope of this fit was also used as a measure of within-session learning. Learning between sessions (an indication of memory consolidation) was calculated as the percent of high-reward lever presses in the first 1/3 of trials on day D + 1 minus the percent of high-reward choices on day D.

The temporal structure of lever-press responding was assessed using an established measure of the “burstiness” of the arrival times of events.63 This measure quantifies burst responding as the average dissimilarity between adjacent intertrial intervals. It is calculated as follows: a value near 0 indicates a perfectly regular interpress interval, values near 1 indicate a Poisson-distributed series of intervals, and values above 1 indicate that there were bursts. We chose this measure because it is more robust relative to other measures of bursting to slow changes in response rates.63 This was a concern because slow changes within days (eg, due to satiety or fatigue) and between days (eg, potentially due to learning) were present.

Back to Top | Article Outline

3. Results

3.1. Learning rates

A goal of the first phase of training was to develop a preference for the right lever. An initial preference for the right lever was established in all groups by only reinforcing right presses on FR and PR tasks (tasks 1-2, Fig. 1A). No differences between groups were observed, as has been previously reported.51 This indicates that SNL animals are not deficient in initial learning and motivation.

In task 3, the reinforcement density of the left lever increased, making it more profitable (right lever: 1 pellet/press, left lever: 4 pellets/press). Assessment of lever preference on day 1 of task 3 indicated that animals expressed a strong preference for the previously reinforced (right) lever (mean preference for high-reward lever: naive = 90% of trials, sham = 93% of trials, SNL = 96% of trials, 1-way ANOVA, F(2,40) = 0.73, P = 0.49). All animals learned to prefer the previously nonreinforced, but now high reinforcement density, left lever over subsequent sessions.

Learning rates were quantified as the slope of the least-squares line fitting the session-by-performance data in Figure 2A. Analysis was restricted to sessions 1 to 8 because these sessions did not involve food restriction (food restriction began on session 9). Spinal nerve ligation rats were slower to adapt to the new contingency than were sham and naive rats (Fig. 2B, 1-way ANOVA F(2,40) = 5.79, P = 0.006, ω2 = 0.18).

We hypothesized that some high-reward (ie, left lever) selections from SNL animals resulted from increased response variability. To investigate this, learning curves were recomputed by only coding high-reward selections that occurred on ≥3 consecutive trials, an approach that eliminated infrequent and unsustained high-reward choices. As with the original % high-reward press measure, this measure also revealed clear differences between SNL and control groups (Fig. 2C, 1-way ANOVA F(2,40) = 8.82, P = 0.0007, ω2 = 0.27). Furthermore, the effect size was considerably higher using this measure (ω2 = 0.27 vs ω2 = 0.18), an observation that further suggests that SNL animals harbored a strong preference for the previously preferred (ie, right) lever. These effects did not seem to result from deficits in motor activity or motivation because SNL animals responded at similar rates (Fig. 3A, 1-way ANOVA F(2,40) = 0.30, P = 0.74) and completed as many trials (Fig. 3B, ANOVA, F(2,40) = 0.67, P = 0.52) as animals in the naive and sham groups.

Figure 3

Figure 3

It was conceivable that some between-group differences were due to an antiallodynic effect of food restriction. Contrary to this prediction, mechanical paw withdrawal responses measured in SNL rats before and during food restriction were 4.56 ± 0.64 and 3.86 ± 0.78 g (SEM), respectively (as previously reported51), indicating that no antiallodynic response was present.

Back to Top | Article Outline

3.2. Burst-like responses in spinal nerve ligation animals

It was conceivable that the temporal pattern of lever-press responding differed between SNL and naive or sham-operated animals. We hypothesized that poor performance in SNL animals could be due to intermittent disengagement from the task, possibly due to distraction or fatigue and the corresponding need for rest. Specifically, we predicted that SNL animals would respond with short bouts of lever presses, followed by long delays. Response sequences consisting of bouts of activity, followed by refractory periods are described as “bursts,” and various metrics have been used to quantify bursting.61 We predicted that the degree of bursting would correlate negatively with learning and be highest in SNL rats (ie, more burst responses in individual rats would correspond to slower learning). Bursting was quantified using the local variance metric.63 Local variance measures the extent to which patterns of intertrial intervals violate a Poisson random process. This measure was originally developed to measure burst activity in neurons (Methods and Fig. 4A). According to this measure, lever presses for SNL rats, but not sham or naive, occurred in bursts on sessions 1 to 8 (Figs. 4B and C 1-way ANOVA, P d1-4 = 0.04, P d5-8 = 0.006, P d10-13 = 0.36. Bonferroni-Holm). Contrary to our hypothesis, local variance was positively correlated with performance on each session in task 3 (Fig. 4D. 1-sample t test, P naive = 0.10, P sham = 0.48, P SNL = 0.007, Bonferroni-Holm), suggesting that bursting increased with improved learning in SNL rats.

Figure 4

Figure 4

Back to Top | Article Outline

3.3. Within-session learning

Impaired associative learning could result from a reduced ability to learn from feedback within a training session. To investigate this possibility, within-session learning was quantified as the slope of a linear fit of trial number to each trial's response (high-reward response coded as 1, low-reward response coded as 0). Thus, this slope indicates that the rate of preference for the high-reward arm changes per trial. Results from this analysis are presented in Figure 5A. No difference in slopes were observed between groups at any time point (Fig. 5B, 1-way ANOVA, all uncorrected P values >0.5). All groups exhibited positive (>0 slope) within-session learning rates during sessions 1 to 4 (Student t test, P < 0.003, Bonferroni-Holm), but not on days 5 to 8 (P > 0.1) or 10 to 13 (P > 0.9). This indicates that in the early trials after a contingency shift, within-session performance improvement occurred comparably across all groups. This finding suggests that SNL rats may not be impaired in their capacity to learn from immediate feedback.

Figure 5

Figure 5

Back to Top | Article Outline

3.4. Between-session learning: memory consolidation and memory retrieval

Associative learning deficits could arise from impaired ability to consolidate memories during posttraining rest or an impaired capacity to retrieve memories on subsequent training days. Under this hypothesis, memory performance was quantified as the difference in performance on adjacent training sessions. This was calculated as the difference in the percent of high-reward responding during the past 17 trials on day X to the first 17 trials on day X + 1. Analysis of this value, which was termed “consolidation,” revealed no group differences at any time point (Figs. 5C and D, 1-way ANOVA, all uncorrected P values >0.5). The consolidation measure was <0 for each group for days 1 to 4 (t test, P < 0.04, Bonferroni-Holm), but was not significantly different from zero on days 5 to 8 (P > 0.2) and 10 to 13 (P > 0.7).

Back to Top | Article Outline

3.5. Variable-ratio performance

After food restriction (sessions 9-13 of task 3), all groups expressed an equivalent and strong preference for the high-reward (left) lever (Figs. 2B and C). The preference for the left lever was then reduced through training on a 2-lever FR (task 4) and cued response task (task 5) that did not favor either lever for reward contingency or magnitude (Methods). Rats were then conditioned on a variable-ratio (VR) schedule (task 6) that favored the left lever (12 pellets per 10 presses) over the right lever (8 pellets per 10 presses), allowing assessment of the preference for the initially preferred (familiar) right lever (trained in tasks 1 and 2). SNL animals did not learn the new contingency because they retained their preference for the initially preferred right lever (Fig. 6. t test, P SNL = 0.52, P naive = 0.006, P sham = 0.01, Bonferroni–Holm correction) despite the continued loss of food reward. There was a main effect of group (1-way ANOVA F(2,40) = 6.67, P = 0.003, ω2 = 0.21) although post hoc between-group comparisons (Tukey–Kramer) only yielded a significant difference between SNL and naive rats.

Figure 6

Figure 6

Back to Top | Article Outline

4. Discussion

The present study was conducted in matched groups of rats under controlled conditions to identify learning deficits associated with chronic neuropathic pain. We showed that SNL animals retain the ability to learn various reinforcement contingencies; however, in a choice task (task 3), they were significantly slower than controls at adapting to new and more profitable behaviors. Analysis of the temporal pattern of lever presses revealed that SNL animals used a unique strategy relative to controls, whereby they responded in bursts of presses. This pattern of response may indicate a coping strategy because burst responding correlated positively with better performance. In a variable-ratio task (task 6), SNL animals were unable to learn a new association and perseverated on the previously learned high-reward lever at a higher rate than naive rats. No deficits in initial learning (tasks 1, 2), within-session learning (task 3), memory consolidation (task 3), or motor ability were seen. These data indicate that chronic pain engendered a selective impairment in cognitive flexibility, with SNL rats responding in bursts of lever presses, followed by longer delays, a pattern that may represent a coping strategy.

When first trained to associate a lever with reward, naive, sham, and SNL animals learn at similar rates (tasks 1 and 2, and as previously reported in Ref. 51), suggesting no deficit in learning the initial association and a specificity for impairment when a new association was introduced. The lack of an initial learning deficit was surprising because impaired learning and memory have been reported in a number of animal models of pain (ie, neuropathic, inflammatory) in tasks such as the Morris water maze27,36,37,69 (although see Ref. 46), novel object recognition,34,41,43,46,56,64,69 and Y-maze23,45,69 (although see Ref. 22). This lack of impairment in SNL animals during the acquisition phase of the task (tasks 1 and 2) may result from the high salience of food reward overcoming deficits in attention produced by chronic pain.16,35 Alternatively, tasks 1 and 2 may represent simpler associative learning behaviors, producing a ceiling effect, whereby all groups learn at high rates.

When reinforcement contingencies were altered, SNL animals were slower than sham and naive controls at acquiring new and optimal behavioral strategies. One factor contributing to this effect may be a reduced capacity of SNL animals to sustain an optimal response strategy. This is suggested by the observation that SNL animals intermittently chose the optimal response, but reverted to the suboptimal and previously profitable lever (Fig. 2C). We suggest that these results are analogous to those found in patients with pain tested in the WCST. When the discriminative stimulus for card matching is switched in the WCST of cognitive flexibility,25 patients with pain persist longer on the previous rule than healthy controls28,65,68 (although this did not reach significance in Ref. 48). The preference of the SNL animals for the previously profitable lever is consistent with the perseverative behavior of patients with pain and suggests translational validity of this task for cognitive flexibility. The data also suggest a causal relationship between chronic pain and cognitive impairment. We note that the mechanisms whereby chronic pain impacts cognitive processing will need further scrutiny to assess possible contributions of pain-related states such as anxiety.

Interestingly, when all animals were food-restricted (Sessions 9-13 of task 3), the learning rate for SNL animals became indistinguishable from naive and sham rats. This could be due to food restriction enhancing motivation in SNL animals; however, our data show no evidence for this because response latencies and the number of completed trials did not differ between groups. Previous reports using rats have also found no motivation deficits for reward under pain conditions (including these SNL animals),26,51 although motivational deficits have been seen in a mouse model.60 Another possibility is that food restriction induced analgesia, which has been observed in an acute and chronic inflammatory pain state,24 although food restriction in our study was very mild by comparison. Further evidence that analgesia due to food restriction was not a major factor is that other studies using similar food-restriction protocols have reported effects of pain on attention,26,53 learning and memory,10,39 and decision making.54 In addition, mechanical paw withdrawal responses measured in SNL rats before and during food restriction were 4.56 ± 0.64 and 3.86 ± 0.78 g, respectively (as previously reported51), demonstrating no antiallodynic effects of our food-restriction protocol. Furthermore, SNL-induced tactile allodynia did not change from day 14 to day 120 post-SNL.51 A limitation of this study design is that all animals were food-restricted on the ninth session, so we were unable to determine whether food restriction was the cause of this recovery, or whether this was the threshold number of training sessions required for SNL rats to fully acquire the new rule.

Findings from previous experiments support the observation that animal models of chronic pain have reduced executive function. For example, rodent models of chronic neuropathic pain exhibit reduced performance in a bowl-digging test of attentional set shifting37,38 and platform relocation in the Morris water maze.36,46 The choice task (task 3) used in the present study had the advantage of allowing further investigation of the nature of the deficit, such as learning across training sessions and the temporal dynamics of responding. The observation of burst-responding in SNL animals indicates that these animals responded reliably for short intervals, followed by periods of inactivity. This could result from animals becoming intermittently distracted or disengaged. This interpretation is supported by previously reported attentional impairments in animal models of pain7,26,41,43,53,64 and patients with chronic pain.15,47,52 It was therefore surprising that burst-responding in SNL animals correlated positively with learning, suggesting that burst-responding could be an adaptive mechanism for overcoming pain-induced cognitive impairment. To the best of our knowledge, this is the first time that an alternative behavioral strategy to a cognitive task has been described in a pain model as a potential compensatory mechanism.

The variable-ratio task studied here is similar to the rodent gambling task, a task in which animal models of pain prefer levers with infrequent high rewards that yield a lower overall return relative to levers with frequent but low-reward amounts.29,54 Therefore, it was surprising that SNL rats in task 6 avoided the high-reward density lever, given that pain should lead to a preference for infrequent and high rewards. This could be related to the type of pain and chronicity of the pain because the previous gambling task studies used an inflammatory pain model,29,54 whereas this was neuropathic. From the perspective of cognitive flexibility, the present data and data from the rodent gambling task are consistent with SNL rats being impaired in shifting preference in the face of environmental pressure. Although rodent and human gambling tasks are often interpreted as tests of probabilistic decision making, many tasks demand that subjects overcome the initial preference for the high-reward and low overall return option. Thus, subjects with chronic pain may not be impaired in probabilistic behaviors per se, but in their capacity to adapt to and overcome well-learned associations.

In summary, these data indicate that animals with chronic neuropathic pain exhibit long-lasting impairment in their capacity to shift away from previously learned associations. These effects were not due to reduced motivation or physical capacity to respond, and therefore represent a selective impact of chronic pain on cognitive flexibility. Furthermore, analysis of the temporal pattern of presses indicated that SNL animals responded in bursts. This unique response pattern may reflect an adaptive response for coping with chronic neuropathic pain. Whether similar results would occur with other types of pain remain to be explored. The reduced capacity of SNL rats to adapt to changing contingencies suggests that chronic pain drives a preference for habitual responses. Habitual responding occurs during stress, whereby behavioral control is shifted away from cortical structures to the dorsal striatum.8,58 Shifts from goal directed to habitual behaviors have not previously been explored in chronic pain. Habitual responding may be a contributing factor to depression, posttraumatic stress disorder, and addiction.59 A shift to habitual responses in chronic pain could contribute to these comorbidities. Indeed, chronic pain is strongly associated with stress,66 and stress and pain are associated with disrupted dopamine function1,9,55 and reduced prefrontal volume,4,5 changes that could negatively impact executive function. Interestingly, chronic pain and stress may have opposing effects on synaptic reorganization within frontal circuits.40,42,61 Taken together, our results suggest that chronic pain encourages a preference for familiar associations and reduces cognitive flexibility. Cognitive flexibility plays a key role in goal-directed behaviors, problem solving, and adapting to changing environments. Thus, reduced cognitive flexibility could have a far-reaching negative impact on the lives of patients with chronic pain.

Back to Top | Article Outline

5. Summary

Rats with chronic neuropathic pain were impaired relative to controls in learning optimal behavioral strategies in the face of changing environmental demands. This suggests that chronic pain disrupts executive functions such as adaptive learning and cognitive flexibility. Specific changes in behavior patterns demonstrated the emergence of a novel learning strategy under conditions of chronic pain. Dysregulation of cognitive flexibility adds an important new dimension to the understanding of the pathophysiological impact of chronic pain.

Back to Top | Article Outline

Conflict of interest statement

J.M. Witkin, D.L. McKinzie, and S.D. Gleason are employees of Eli Lilly & Company. The remaining authors have no conflict of interest to declare.

This work was partially funded by DA034975 from the NIH-NIDA.

Back to Top | Article Outline


The authors thank Eli Lilly and Company for their collaborative support and partial funding of this work. The authors thank the surgical team at Covance (Greenfield, IN) for generation of the SNL and sham rats. The authors thank Rachel Samson, Anne Smith, and Brian McElroy for their technical assistance.

Author contributions: F. Porreca, E. Navratilova, A. Okun, J.M. Witkin, and D.L. McKinzie designed the study. J.M. Witkin, A. Okun, D.L.McKinzie, O. Husain, and S.D. Gleason collected the data. S.L. Cowen performed the data analysis. S.L. Cowen, C.E. Phelps, E. Navratilova, J.M. Witkin, and F. Porreca wrote the manuscript.

Back to Top | Article Outline


[1]. Albrecht DS, MacKie PJ, Kareken DA, Hutchins GD, Chumin EJ, Christian BT, Yoder KK. Differential dopamine function in fibromyalgia. Brain Imaging Behav 2016;10:829–839.
[2]. Alvarado S, Tajerian M, Millecamps M, Suderman M, Stone LS, Szyf M. Peripheral nerve injury is accompanied by chronic transcriptome-wide changes in the mouse prefrontal cortex. Mol Pain 2013;9:21.
[3]. Alvarez JA, Emory E. Executive function and the frontal lobes: a meta-analytic review. Neuropsychol Rev 2006;16:17–42.
[4]. Ansell EB, Rando K, Tuit K, Guarnaccia J, Sinha R. Cumulative adversity and smaller gray matter volume in medial prefrontal, anterior cingulate, and insula regions. Biol Psychiatry 2012;72:57–64.
[5]. Apkarian AV, Sosa Y, Sonty S, Levy RM, Harden RN, Parrish TB, Gitelman DR. Chronic back pain is associated with decreased prefrontal and thalamic gray matter density. J Neurosci 2004;24:10410–10415.
[6]. Arnett PA, Rao SM, Bernardin L, Grafman J, Yetkin FZ, Lobeck L. Relationship between frontal lobe lesions and Wisconsin Card Sorting Test performance in patients with multiple sclerosis. Neurology 1994;44(3 pt 1):420–5.
[7]. Boyette-Davis JA, Thompson CD, Fuchs PN. Alterations in attentional mechanisms in response to acute inflammatory pain and morphine administration. Neuroscience 2008;151:558–63.
[8]. Braun S, Hauber W. Acute stressor effects on goal-directed action in rats. Learn Mem 2013;20:700–9.
[9]. Butts KA, Weinberg J, Young AH, Phillips AG. Glucocorticoid receptors in the prefrontal cortex regulate stress-evoked dopamine efflux and aspects of executive function. Proc Natl Acad Sci U S A 2011;108:18459–64.
[10]. Cain CK, Francis JM, Plone MA, Emerich DF, Lindner MD. Pain-related disability and effects of chronic morphine in the adjuvant-induced arthritis model of chronic pain. Physiol Behav 1997;62:199–205.
[11]. Cardoso-Cruz H, Lima D, Galhardo V. Impaired spatial memory performance in a rat model of neuropathic pain is associated with reduced hippocampus-prefrontal cortex connectivity. J Neurosci 2013;33:2465–80.
[12]. Dajani DR, Uddin LQ. Demystifying cognitive flexibility: implications for clinical and developmental neuroscience. Trends Neurosci 2015;38:571–8.
[13]. Davis JC, Marra CA, Najafzadeh M, Liu-Ambrose T. The independent contribution of executive functions to health related quality of life in older women. BMC Geriatr 2010;10:16.
[14]. Devoize L, Alvarez P, Monconduit L, Dallel R. Representation of dynamic mechanical allodynia in the ventral medial prefrontal cortex of trigeminal neuropathic rats. Eur J Pain 2011;15:676–82.
[15]. Eccleston C. Chronic pain and attention: a cognitive approach. Br J Clin Psychol 1994;33(pt 4):535–47.
[16]. Eccleston C, Crombez G. Pain demands attention: a cognitive-affective model of the interruptive function of pain. Psychol Bull 1999;125:356–66.
[17]. Engel de Abreu PM, Abreu N, Nikaedo CC, Puglisi ML, Tourinho CJ, Miranda MC, Befi-Lopes DM, Bueno OF, Martin R. Executive functioning and reading achievement in school: a study of Brazilian children assessed by their teachers as “poor readers”. Front Psychol 2014;5:550.
[18]. Fritz HC, McAuley JH, Wittfeld K, Hegenscheid K, Schmidt CO, Langner S, Lotze M. Chronic back pain is associated with decreased prefrontal and anterior insular gray matter: results from a population-based cohort study. J Pain 2016;17:111–18.
[19]. Genet JJ, Siemer M. Flexible control in processing affective and non-affective material predicts individual differences in trait resilience. Cogn Emot 2011;25:380–8.
[20]. Goldstein B, Obrzut JE, John C, Ledakis G, Armstrong CL. The impact of frontal and non-frontal brain tumor lesions on Wisconsin Card Sorting Test performance. Brain Cogn 2004;54:110–16.
[21]. Grant DA, Berg EA. A behavioral analysis of degree of reinforcement and ease of shifting to new responses in a Weigl-type card-sorting problem. J Exp Psychol 1948;38:404–11.
[22]. Grégoire S, Michaud V, Chapuy E, Eschalier A, Ardid D. Study of emotional and cognitive impairments in mononeuropathic rats: effect of duloxetine and gabapentin. PAIN 2012;153:1657–63.
[23]. Grégoire S, Wattiez AS, Etienne M, Marchand F, Ardid D. Monoarthritis-induced emotional and cognitive impairments in rats are sensitive to low systemic doses or intra-amygdala injections of morphine. Eur J Pharmacol 2014;735:1–9.
[24]. Hargraves WA, Hentall ID. Analgesic effects of dietary caloric restriction in adult mice. PAIN 2005;114:455–61.
[25]. Heaton RK, Chelune GJ, Talley JL, Kay GG, Curtis G. Wisconsin card sorting test (WCST) manual revised and expanded. Odessa: Psychological Assessment Resources Inc, 1993.
[26]. Higgins GA, Silenieks LB, Van Niekerk A, Desnoyer J, Patrick A, Lau W, Thevarkunnel S. Enduring attentional deficits in rats treated with a peripheral nerve injury. Behav Brain Res 2015;286:347–55.
[27]. Hu Y, Yang J, Wang Y, Li W. Amitriptyline rather than lornoxicam ameliorates neuropathic pain-induced deficits in abilities of spatial learning and memory. Eur J Anaesthesiol 2010;27:162–8.
[28]. Indart S, Hugon J, Guillausseau PJ, Gilbert A, Dumurgier J, Paquet C, Sène D. Impact of pain on cognitive functions in primary Sjögren syndrome with small fiber neuropathy: 10 cases and a literature review. Medicine (Baltimore) 2017;96:e6384.
[29]. Ji G, Sun H, Fu Y, Li Z, Pais-Vieira M, Galhardo V, Neugebauer V. Cognitive impairment in pain through amygdala-driven prefrontal cortical deactivation. J Neurosci 2010;30:5451–64.
[30]. Karp JF, Reynolds CF, Butters MA, Dew MA, Mazumdar S, Begley AE, Lenze E, Weiner DK. The relationship between pain and mental flexibility in older adult pain clinic patients. Pain Med 2006;7:444–52.
[31]. Kelly CJ, Huang M, Meltzer H, Martina M. Reduced glutamatergic currents and dendritic branching of layer 5 pyramidal cells contribute to medial prefrontal cortex deactivation in a rat model of neuropathic pain. Front Cell Neurosci 2016;10:133.
[32]. Kesner RP, Churchwell JC. An analysis of rat prefrontal cortex in mediating executive function. Neurobiol Learn Mem 2011;96:417–31.
[33]. Kim SH, Chung JM. An experimental model for peripheral neuropathy produced by segmental spinal nerve ligation in the rat. PAIN 1992;50:355–63.
[34]. Kodama D, Ono H, Tanabe M. Increased hippocampal glycine uptake and cognitive dysfunction after peripheral nerve injury. PAIN 2011;152:809–817.
[35]. Legrain V, Damme SV, Eccleston C, Davis KD, Seminowicz DA, Crombez G. A neurocognitive model of attention to pain: behavioral and neuroimaging evidence. PAIN 2009;144:230–2.
[36]. Leite-Almeida H, Almeida-Torres L, Mesquita AR, Pertovaara A, Sousa N, Cerqueira JJ, Almeida A. The impact of age on emotional and cognitive behaviours triggered by experimental neuropathy in rats. PAIN 2009;144:57–65.
[37]. Leite-Almeida H, Cerqueira JJ, Wei H, Ribeiro-Costa N, Anjos-Martins H, Sousa N, Pertovaara A, Almeida A. Differential effects of left/right neuropathy on rats' anxiety and cognitive behavior. PAIN 2012;153:2218–25.
[38]. Leite-Almeida H, Guimarães MR, Cerqueira JJ, Ribeiro-Costa N, Anjos-Martins H, Sousa N, Almeida A. Asymmetric c-fos expression in the ventral orbital cortex is associated with impaired reversal learning in a right-sided neuropathy. Mol Pain 2014;10:41.
[39]. Lindner MD, Plone MA, Francis JM, Cain CK. Chronic morphine reduces pain-related disability in a rodent model of chronic, inflammatory pain. Exp Clin Psychopharmacol 1999;7:187–97.
[40]. Liston C, Miller MM, Goldwater DS, Radley JJ, Rocher AB, Hof PR, Morrison JH, McEwen BS. Stress-induced alterations in prefrontal cortical dendritic morphology predict selective impairments in perceptual attentional set-shifting. J Neurosci 2006;26:7870–4.
[41]. Low LA, Millecamps M, Seminowicz DA, Naso L, Thompson SJ, Stone LS, Bushnell MC. Nerve injury causes long-term attentional deficits in rats. Neurosci Lett 2012;529:103–7.
[42]. Metz AE, Yau HJ, Centeno MV, Apkarian AV, Martina M. Morphological and functional reorganization of rat medial prefrontal cortex in neuropathic pain. Proc Natl Acad Sci U S A 2009;106:2423–8.
[43]. Millecamps M, Etienne M, Jourdan D, Eschalier A, Ardid D. Decrease in non-selective, non-sustained attention induced by a chronic visceral inflammatory state as a new pain evaluation in rats. PAIN 2004;109:214–24.
[44]. Milner B. Effects of different brain lesions on card sorting: the role of the frontal lobes. Arch Neurol 1963;9:90–100.
[45]. Morel V, Etienne M, Wattiez AS, Dupuis A, Privat AM, Chalus M, Eschalier A, Daulhac L, Pickering G. Memantine, a promising drug for the prevention of neuropathic pain in rat. Eur J Pharmacol 2013;721:382–90.
[46]. Moriarty O, Gorman CL, McGowan F, Ford GK, Roche M, Thompson K, Dockery P, McGuire BE, Finn DP. Impaired recognition memory and cognitive flexibility in the rat L5–L6 spinal nerve ligation model of neuropathic pain. Scand J Pain 2016;10:61–73.
[47]. Moriarty O, McGuire BE, Finn DP. The effect of pain on cognitive function: a review of clinical and preclinical research. Prog Neurobiol 2011;93:385–404.
[48]. Moriarty O, Ruane N, O'Gorman D, Maharaj CH, Mitchell C, Sarma KM, Finn DP, McGuire BE. Cognitive impairment in patients with chronic neuropathic or radicular pain: an interaction of pain and age. Front Behav Neurosci 2017;11:100.
[49]. Mukhopadhyay P, Dutt A, Kumar Das S, Basu A, Hazra A, Dhibar T, Roy T. Identification of neuroanatomical substrates of set-shifting ability: evidence from patients with focal brain lesions. Prog Brain Res 2008;168:95–104.
[50]. Nyhus E, Barceló F. The Wisconsin Card Sorting Test and the cognitive assessment of prefrontal executive functions: a critical update. Brain Cogn 2009;71:437–51.
[51]. Okun A, McKinzie DL, Witkin JM, Remeniuk B, Husein O, Gleason SD, Oyarzo J, Navratilova E, McElroy B, Cowen S, Kennedy JD, Porreca F. Hedonic and motivational responses to food reward are unchanged in rats with neuropathic pain. PAIN 2016;157:2731–8.
[52]. Oosterman J, Derksen LC, van Wijck AJ, Kessels RP, Veldhuijzen DS. Executive and attentional functions in chronic pain: does performance decrease with increasing task load? Pain Res Manag 2012;17:159–65.
[53]. Pais-Vieira M, Lima D, Galhardo V. Sustained attention deficits in rats with chronic inflammatory pain. Neurosci Lett 2009;463:98–102.
[54]. Pais-Vieira M, Mendes-Pinto MM, Lima D, Galhardo V. Cognitive impairment of prefrontal-dependent decision-making in rats after the onset of chronic pain. Neuroscience 2009;161:671–9.
[55]. Pani L, Porcella A, Gessa GL. The role of stress in the pathophysiology of the dopaminergic system. Mol Psychiatry 2000;5:14–21.
[56]. Ren WJ, Liu Y, Zhou LJ, Li W, Zhong Y, Pang RP, Xin WJ, Wei XH, Wang J, Zhu HQ, Wu CY, Qin ZH, Liu G, Liu XG. Peripheral nerve injury leads to working memory deficits and dysfunction of the hippocampus by upregulation of TNF-α in rodents. Neuropsychopharmacology 2011;36:979–92.
[57]. Schiltenwolf M, Akbar M, Neubauer E, Gantz S, Flor H, Hug A, Wang H. The cognitive impact of chronic low back pain: positive effect of multidisciplinary pain therapy. Scand J Pain 2017;17:273–8.
[58]. Schwabe L, Tegenthoff M, Höffken O, Wolf OT. Mineralocorticoid receptor blockade prevents stress-induced modulation of multiple memory systems in the human brain. Biol Psychiatry 2013;74:801–8.
[59]. Schwabe L, Wolf OT. Stress and multiple memory systems: from “thinking” to “doing”. Trends Cogn Sci 2013;17:60–8.
[60]. Schwartz N, Temkin P, Jurado S, Lim BK, Heifets BD, Polepalli JS, Malenka RC. Chronic pain. Decreased motivation during chronic pain requires long-term depression in the nucleus accumbens. Science 2014;345:535–42.
[61]. Seib LM, Wellman CL. Daily injections alter spine density in rat medial prefrontal cortex. Neurosci Lett 2003;337:29–32.
[62]. Seminowicz DA, Laferriere AL, Millecamps M, Yu JS, Coderre TJ, Bushnell MC. MRI structural brain changes associated with sensory and emotional function in a rat model of long-term neuropathic pain. Neuroimage 2009;47:1007–14.
[63]. Shinomoto S, Kim H, Shimokawa T, Matsuno N, Funahashi S, Shima K, Fujita I, Tamura H, Doi T, Kawano K, Inaba N, Fukushima K, Kurkin S, Kurata K, Taira M, Tsutsui K, Komatsu H, Ogawa T, Koida K, Tanji J, Toyama K. Relating neuronal firing patterns to functional differentiation of cerebral cortex. PLoS Comput Biol 2009;5:e1000433.
[64]. Suto T, Eisenach JC, Hayashida K. Peripheral nerve injury and gabapentin, but not their combination, impair attentional behavior via direct effects on noradrenergic signaling in the brain. PAIN 2014;155:1935–42.
[65]. Tamburin S, Maier A, Schiff S, Lauriola MF, Di Rosa E, Zanette G, Mapelli D. Cognition and emotional decision-making in chronic low back pain: an ERPs study during Iowa gambling task. Front Psychol 2014;5:1350.
[66]. Ulrich-Lai YM, Xie W, Meij JT, Dolgas CM, Yu L, Herman JP. Limbic and HPA axis function in an animal model of chronic neuropathic pain. Physiol Behav 2006;88:67–76.
[67]. van der Leeuw G, Eggermont LH, Shi L, Milberg WP, Gross AL, Hausdorff JM, Bean JF, Leveille SG. Pain and cognitive function among older adults living in the community. J Gerontol A Biol Sci Med Sci 2016;71:398–405.
[68]. Verdejo-García A, López-Torrecillas F, Calandre EP, Delgado-Rodríguez A, Bechara A. Executive function and decision-making in women with fibromyalgia. Arch Clin Neuropsychol 2009;24:113–22.
[69]. Wu J, Zhao Z, Sabirzhanov B, Stoica BA, Kumar A, Luo T, Skovira J, Faden AI. Spinal cord injury causes brain inflammation associated with cognitive and affective changes: role of cell cycle pathways. J Neurosci 2014;34:10989–1006.
[70]. Yuan P, Raz N. Prefrontal cortex and executive functions in healthy adults: a meta-analysis of structural neuroimaging studies. Neurosci Biobehav Rev 2014;42:180–92.

Neuropathic pain; Executive function; Associative learning; Decision making

© 2018 International Association for the Study of Pain