- Email: [email protected]
Effects of optimism on gambling in the rat slot machine task
Behavioural Brain Research 300 (2016) 97–105
Contents lists available at ScienceDirect
Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr
Research report
Effects of optimism on gambling in the rat slot machine task Dominik Rafa 1 , Jakub Kregiel 1 , Piotr Popik, Rafal Rygula ∗ Affective Cognitive Neuroscience Lab, Department of Behavioral Neuroscience and Drug Development, Institute of Pharmacology Polish Academy of Sciences, 12 Smetna Street, 31-343 Krakow, Poland
h i g h l i g h t s • • • •
We identified rats displaying ‘pessimistic’ and ‘optimistic’ traits. We determined how these traits affect behavior in the rat Slot Machine Task. ‘Optimistic’ rats were more likely to gamble in the hopeless ‘clear loss’ situation. We demonstrated interrelation between optimism and gambling in an animal model.
a r t i c l e
i n f o
Article history: Received 21 October 2015 Received in revised form 30 November 2015 Accepted 11 December 2015 Available online 14 December 2015 Keywords: Rat Gambling Ambiguous-cue interpretation Pessimism Optimism Cognitive judgment bias
a b s t r a c t Although gambling disorder is a serious social problem in modern societies, information about the behavioral traits that could determine vulnerability to this psychopathology is still scarce. In this study, we used a recently developed ambiguous-cue interpretation (ACI) paradigm to investigate whether ‘optimism’ and ‘pessimism’ as behavioral traits may determine the gambling-like behavior of rodents. In a series of ACI tests (cognitive bias screening), we identified rats that displayed ‘pessimistic’ and ‘optimistic’ traits. Subsequently, using the rat slot machine task (rSMT), we investigated if the ‘optimistic’/‘pessimistic’ traits could determine the crucial feature of gambling-like behavior that has been investigated in rats and humans: the interpretation of ‘near-miss’ outcomes as a positive (i.e., win) situation. We found that ‘optimists’ did not interpret ‘near-miss’, ‘near loss’, or ‘clear win’ as win trials more often than their ‘pessimistic’ conspecifics; however, the ‘optimists’ were statistically more likely to reach for a reward in the hopeless ‘clear loss’ situation. This agrees with human studies and provides a platform for modeling interactions between behavioral traits and gambling in animals. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Cognitive theories of psychiatric disorders have articulated the mediating role of cognitive biases, distortions in reasoning, and errors in judgment in the phenomenology of behavioral addictions, such as gambling disorder [1–3]. One of the errors in reasoning that may be critically involved in gambling disorder is cognitive judgment bias that is manifested by excessive optimism. According to Carver and Scheier [4], the generalized positive outcome expectancies of optimists result persistence in attempting to accomplish goals in the face of adversity, whereas the generalized negative outcome expectancies of pessimists result in withdrawal. Thus, the benefits of optimism may arise primarily in situations in which this persistence is rewarded. Unfortunately, not all situations and
∗ Corresponding author. Fax: +48 12 6374500. E-mail address: [email protected] (R. Rygula). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.bbr.2015.12.013 0166-4328/© 2015 Elsevier B.V. All rights reserved.
tasks have positive outcomes. Gambling is one important domain in which persistence is unlikely to be consistently rewarded and in which optimism may be a liability [5]. In agreement with this assumption, recent research on optimism and gambling has found that optimists had more positive expectations for gambling than did pessimists and were less likely to reduce their betting after poor outcomes [5]. Despite these findings, there is still a dearth of information concerning the interaction of optimism and gambling in both humans and rodents, and despite the important role of cognitive biases in psychopathologies, they have not been the subject of much preclinical behavioral research. Although several research groups over the past decade have attempted to measure the effects of various behavioral and pharmacological manipulations on ‘optimistic’ and ‘pessimistic’ judgment biases [6–13], almost none of them have investigated ‘optimism’/‘pessimism’ as an enduring and stable behavioral trait that could be used to evaluate its implications for psychiatric conditions, such as gambling disorder.
98
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
We have recently shown that, similar to humans, the cognitive judgment bias of animals has components of both enduring traits and transient states [14]. A trait is the stable level of an individual’s ‘pessimism’/‘optimism’ that is generally experienced, whereas a state represents the valence of cognitive judgment bias that may change based upon the situation or contextual factors. We also have identified a relationship between cognitive judgment bias as a behavioral trait and vulnerability of animals to stress-induced anhedonia [14] as well as a correlation between the level of optimism and motivation [15]. In the present study, we used multiple ambiguous-cue interpretation (ACI) tests [10,11,13] to isolate two groups of rats that consistently differed in their cognitive judgment bias over time [14,15]. These two groups of ‘pessimistic’ and ‘optimistic’ rats were subsequently trained and tested for the propensity to gamble in the rat slot machine task (rSMT) [16]. In this behavioral paradigm subjects respond to a series of three flashing lights, loosely analogous to the wheels of a slot machine, causing the lights to set to ‘ON’ or ‘OFF’. A winning outcome is signaled if all three lights were illuminated. At the end of each trial, rats chose between responding on the ‘collect’ lever, resulting in reward on win trials, but a time penalty on loss trials, or starting a new trial. Recent studies using this task revealed important role of dopamine in pathological gambling [16–18]. We hypothesized that, similar to humans, rats display different propensities to gamble in association with traits of ‘optimism’ and ‘pessimism.’
2. Methods 2.1. Ethics statement All described experimental procedures were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Committee for Ethics in Animal Experiments at the Institute of Pharmacology Polish Academy of Sciences. 2.2. Subjects and housing We used 24 male Sprague Dawley rats (Charles River, Germany) that weighed between 175 and 200 g upon arrival. The animals were housed in groups of five, in a temperature (21 ± 1 ◦ C) and humidity (40–50%) controlled room under a 12/12 h dark/light cycle (lights on at 07:00 h). The animals were habituated to the housing conditions and experimental facility for two weeks after arrival and before the start of experiments. In all experiments, the animals received 15–20 g of food per rat per day (standard laboratory chow), which corresponds to mild food-deprivation. The food deprivation began seven days prior to beginning of the training. The rats were weighed once a week. The water was provided ad libitum . The animals were trained and tested during the light phase of the dark/light cycle. The rats were habituated to the experimental room for 30 min prior to training and testing sessions and were tested once daily. 2.3. Ambiguous-cue interpretation test 2.3.1. Apparatus The experiments were conducted in eight computer-controlled, operant conditioning boxes (Med Associates, St. Albans, Vermont, USA). The boxes were equipped with lights, speakers, liquid dispensers (0.1 ml of 5% sucrose solution), electric grid floors, and two retractable levers. The levers were located on both sides of the
liquid dispenser. The experimental protocols were written in Med State notation code (Med Associates). 2.3.2. Behavioral training The experimental training and testing procedures for the ACI paradigm that was used in this study were modified from procedures previously described by Enkel et al. [10] and have been described in detail elsewhere [13,14,19–23]. In brief, initially the animals were trained to press one lever when a ‘positive’ tone (2000 Hz at 75 dB) signaled the availability of a reward (5% sucrose solution) and to press second lever when another, ‘negative’ tone (9000 Hz at 75 dB) signaled a forthcoming punishment (0.5 mA, 10 s). By pressing appropriate levers the animals could either receive a reward or avoid punishment. The tone presentations were separated by 10 s intertrial intervals (ITI) and each training session lasted 30 min. The animals had to fulfill the criteria of at least 90% of accurate responses to the tone signaling reward availability maintained over three consecutive training sessions and, at least 60% of correct punishment-prevention responses maintained over three consecutive training sessions, to proceed to the discrimination training. During the discrimination-training phase, the animals were trained to discriminate between pseudo-randomly presented positive (20) and negative (20) tones by responding to the appropriate levers (as learned in the previous training stages), thereby minimizing punishment and maximizing reward delivery. Each discrimination training session lasted 40 min. The animals had to achieve a minimum of 70% correct responses with each lever, maintained over three consecutive discrimination sessions to be qualified for the ACI testing. As for an animal it is really difficult to learn instrumental avoidance reaction in anticipation of punishment, the criterion at the stage of ‘negative tone’ training was lower than for the ‘positive tone’ training. However, at the stage of discrimination, all animals were trained until they reached equally stable discrimination ratio for both tones—minimum 70%. 2.3.3. Ambiguous-cue testing During the ACI testing sessions the animals were exposed to 20 negative, 20 positive, and 10 ambiguous (5000 Hz at 75 dB) tone presentations. The tones were played in a pseudo-randomized order and were separated by 10 s ITIs. The responses to each tone (positive, ambiguous and negative) during the ACI testing were analyzed as the proportion of the overall number of responses to a given tone. To calculate the cognitive bias index, we subtracted the proportion of negative responses to the ambiguous-cues from the proportion of positive responses to the ambiguous-cue, which resulted in values ranging between −1 and 1. After median split, values above 0.14 indicated an overall positive judgment and ‘optimistic’ interpretation of the ambiguous-cue while the values below 0.14 indicated overall negative judgment and ‘pessimism’. 2.3.4. Cognitive bias screening (CBS) The CBS procedure has been described in detail elsewhere [14]. Briefly, to assess the cognitive judgment bias as a trait, we examined the animals using a series of 10 consecutive ACI tests that were conducted at one-week intervals. Based upon the average cognitive bias index that was obtained from these 10 ACI tests, the rats were divided into two subgroups: ‘optimistic’ and ‘pessimistic’. 2.4. Rat slot machine task test The experimental set-up, training and testing procedures for the rSMT paradigm used in this study were adapted and modified from procedures that were previously described by Winstanley et al. [16].
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
99
Following the CBS screening, the animals were re-trained for the rSMT. To minimize the impact of one task on the other, the animals were re-trained and tested by another experimenter in another experimental room, using different set of operant boxes (Coulbourn Instruments, Whitehall, PA, USA). 2.4.1. Apparatus The experiments were conducted in four computer-controlled, operant conditioning boxes (Coulbourn Instruments, USA) each enclosed within a ventilated sound-attenuating cabinet. Each box was fitted with the loudspeaker, house light, two retractable levers and an array of three nose poke response holes that were positioned 2 cm above a bar floor on one of the walls. A stimulus light was set at the back of each hole. Infrared beams detected nose-poke responses into these apertures. A food tray was located in the middle of the opposite wall with one retractable lever located on each side. One of the levers served as the ‘collect’ lever. By pressing it, the rat indicated a choice that suggested that the animal assumed the trial was the ‘win’ trial. The other retractable lever was a ‘roll’ lever. By pressing it, the rat indicated that the animal chose not to collect the reward but rather to initiate or continue the game. The loudspeaker was located above the food tray. The apparatuses were controlled by Graphic State 3.0 software (Coulbourn Instruments, USA). 2.4.2. Behavioral training All training and testing stages were adapted from Winstanley et al. [16]. In brief, initially the animals were placed in the operant chamber with a house light turned on and with sucrose pellets (45 mg, Bio-Serv, New Jersey, USA) located in the food tray and nose-poke apertures. Each session lasted 30 min. Then, rats were trained to press the levers to earn the food reward under a fixed ratio 1 (FR1) schedule of reinforcement. Only one lever was presented during the session. When the animal had made more than 50 lever presses in a session, training was repeated on the other lever. The order in which the levers were presented (left/right) was counterbalanced between the subjects. Once the animals had learnt to press the levers, the proper training for the rSMT began. 2.4.3. Rat slot machine task Following the completion of successive versions of the slot machine program, which gradually increased in complexity (described in detail by Winstanley et al. [16] and Cocker et al. [18]), the proper rSMT training and testing began. The rat had to press the ‘roll’ lever that triggered its retraction and started the trial by flashing the first hole light at a frequency of 2 Hz (Fig. 1). When the rat nose-poked into the first hole, depending on the schedule, the light inside the hole was either turned ON or turned OFF for the rest of the trial. The ON or OFF status of the hole was accompanied by either a high (15 000 Hz at 68 dB) or low (4000 Hz at 68 dB) acoustic tone, which served as an additional cue. The sounds lasted for 1 s. Following the first hole step, the light in second hole began to flash. When the rat made a nose-poke response into the second hole, as in the previous step, the light inside was either turned ON or turned OFF, depending on the schedule, and a high or low relevant acoustic tone was presented for 1 s. Then, the light inside the third hole began to flash, and the sequence of events (depending on the schedule, light ON/OFF accompanied with a relevant acoustic signal) was repeated, similar to the procedure for the first and second holes. At this moment, the animal was exposed to and, by nose-poking, responded to the series of three stimuli. To increase the number of meaningful trials, we modified the original procedure that was described by Winstanley et al. [16], and reduced the number of possible two and three light pattern outcomes to (ON–ON–OFF or OFF–ON–ON) and (ON–OFF–OFF and OFF–OFF–ON), respectively. In this way each trial could have
Fig. 1. (A) Schematic representation of the experimental schedule and main result of the study. (B) Scheme of the rSMT test session.
only six out of eight possible patterns: ON–ON–ON, ON–ON–OFF, OFF–ON–ON, ON–OFF–OFF, OFF–OFF–ON and OFF–OFF–OFF. This gave one state with three lights ON (a rewarded, clear ‘win’), two near-miss states with two lights ON (ON–ON–OFF or OFF–ON–ON, ‘near miss’), two states with one light ON (ON–OFF–OFF, OFF–OFF–ON, ‘near loss’) and one state with all lights OFF (a clear ‘loss’).
100
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
For analysis, the data from two-light trials ((ON–ON–OFF) and (OFF–ON–ON)) and one-light trials ((ON–OFF–OFF) and (OFF–OFF–ON)) were pooled. The frequency of all six light patterns was predefined in a pseudo-random order and each pattern was presented once per block of six trials so that the animals could obtain a food reward only once in each block. The pseudo-random order of trials in each block was different so that animals could not learn the sequence. The rat was required to interpret the given pattern as a ‘win’ or ‘loss’. This was accomplished at the ‘decision point’ by extending both the ‘roll’ and ‘collect’ levers. On the clear ‘win’ trials, during which all three lights were set to ON, pressing the ‘collect’ lever resulted in a delivery of five sucrose pellets. On any other pattern, with at least one light being turned OFF, responding on the ‘collect’ lever resulted in a penalty of 25 s of time-out, during which the animal could not quickly start a new trial. However, pressing the ‘roll’ lever on any trial cancelled the potential reward or the time-out, so that the rat could immediately start a new trial. The rat’s optimal strategy was to respond with the ‘collect’ lever only on ‘win’ trials to obtain the food reward and to respond with a ‘roll’ lever on every ‘loss’ trial to avoid the time-out and start a new trial. There was no time requirement to respond on the lever or the nose-poke. Rats were trained once a day, five days per week, until they established a stable performance. The animals were deemed to have successfully acquired the task if they completed ≥25 trials per session and made ≤50% collect responses on clear loss (OFF–OFF–OFF) trials, five times in a row. 2.5. Experimental design and behavioral measures The experimental design is schematically presented in Fig. 1. After establishing the ‘optimistic’ and ‘pessimistic’ traits in the individual animals (CBS procedure), the rats were divided into two experimental groups: ‘optimistic’ and ‘pessimistic’ (Fig. 1A). To assess whether the traits of ‘optimism’ and ‘pessimism’ may determine the propensity of the animals to gamble, the rats were re-trained in another set of four experimental boxes (Coulbourn Instruments, USA) and subsequently tested for five consecutive days on the rSMT. After reaching the training criteria (described above), the animals were tested five times on the rSMT. Average results from these five tests were compared between the groups of animals that were previously classified as ‘optimistic’ and ‘pessimistic’. 2.6. Statistics We analyzed the data using SPSS (version 21.0, SPSS Inc., Chicago, IL, United States). The differences in the processing of the experimental tones between the ‘optimists’ and ‘pessimists’ were investigated using a three-way analysis of variance (ANOVA) with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factors of lever (two levels: positive and negative) and tone (three levels: positive, ambiguous and negative). The stability of the trait was analyzed using a two-way ANOVA with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factor of test (10 levels: 1–10). The effects of different light patterns within 2 and 1-light trials on responding to ‘collect’ lever were analyzed using a two-way ANOVA with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factor of the type of light pattern (2 levels: ON–ON–OFF and OFF–ON–ON for 2 light trials, and 2 levels: ON–OFF–OFF and OFF–OFF–ON for 1 light trials). The differences between ‘optimistic’ and ‘pessimistic’ animals in the length of training, ‘optimism’ frequency, numbers of trials finished per experimental session, responding to the ‘collect’ lever at different light patterns in the rSMT and response latencies
Fig. 2. ‘Optimistic’ vs. ‘pessimistic’ animals; results of the cognitive bias screening. (A) The mean ± SEM of the cognitive bias index of the animals that were classified (based on 10 ACI tests) as ‘pessimistic’ (circles, N = 7) vs. ‘optimistic’ (squares, N = 6). A cognitive bias index above 0.14 indicates an overall positive judgment and ‘optimistic’ interpretation of the ambiguous cue. (B) The mean ± SEM of the proportion of positive, (C) negative and (D) omitted responses to the trained and ambiguous tones in the ‘optimistic’ (open circles, N = 6) and ‘pessimistic’ (filled circles, N = 7) rat groups. * indicates significant (p ≤ 0.05) differences between the ‘optimistic’ and ‘pessimistic’ animals.
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
were analyzed separately using t-tests. For pair-wise comparisons, we adjusted the values using Sidak’s correction factor for multiple comparisons [24]. All tests of significance were performed at ˛ = 0.05. We tested the homogeneity of variance using Levene’s test and, for repeated-measures analyses, we confirmed the sphericity using Mauchly’s test. The data are presented as the mean ± SEM.
3. Results Out of the 24 initially trained animals, 24 qualified for CBS using the ACI test; however, after re-training in the rSMT, only 13 reached the training criteria and qualified for the final testing. It is worth noticing that the animals were excluded, because they were unable to successfully acquire the rSMT task by completion of more than 25 trials per session and not because they could not make ≤50% collect responses on clear loss (OFF–OFF–OFF) trials, five times in a row. Out of 11 excluded animals 6 were ‘optimistic’ and 5 were ‘pessimistic’. During the ACI training, the ‘optimists’ reached the criteria of ‘positive’ tone, ‘negative’ tone and discrimination trainings after 13.2 ± 0.1, 30.0 ± 3.7 and 43.2 ± 1.2 training sessions, respectively, whereas the ‘pessimists’ reached the criteria after 13.1 ± 0.14, 27.1 ± 4.2 and 42.1 ± 1.0 training sessions, respectively. We observed no significant differences in the length of ‘positive’ tone training (t(11) = 0.11, NS), ‘negative’ tone training (t(11) = 0.50, NS), discrimination training (t(11) = 0.79, NS) and in the total duration of training (t(11) = 0.68, NS) between the ‘optimists’ (86.4 ± 4.4 training sessions) and ‘pessimists’ (82.3 ± 3.8 training sessions). ‘Optimists’ and ‘pessimists’ did not significantly differ in body weight (t(11) = 0.13, NS).
3.1. ‘Pessimistic’ vs. ‘optimistic’ rats Based on the results of the CBS (Fig. 3A), we divided the animals that qualified for the rSMT into two groups using the median split. These groups were clearly distinctive in their interpretation of the ambiguous-cues over time: ‘optimists’ (N = 6, average (AVG) cognitive bias index >0.14) and ‘pessimists’ (N = 7, AVG cognitive bias index <0.14). Further analysis revealed significant differences in the patterns of lever responses between ‘optimists’ and ‘pessimists’ (lever × tone × cognitive bias interaction (F(2,22) = 11.4, p < 0.001)). The ‘optimists’ responded significantly more often to the positive lever in response to the ambiguous and negative tones tone when compared to ‘pessimists’ (p < 0.002 and p = 0.033, respectively, Fig. 2B) and less often to the negative lever in response to the positive and negative tones (p = 0.002 and p = 0.005, respectively, Fig. 2C). As shown in Fig. 2D, ‘optimists’ and ‘pessimists’ did not differ in the numbers of omitted trials (no significant effects of cognitive bias (F(1,11) = 0.002, NS) or in the cognitive bias × tone interaction (F(2,22) = 1.87, NS)). ‘Optimistic’ animals showed significantly higher average frequencies of ‘optimism’ (number of tests when the cognitive bias index of an individual animal was higher than zero, out of the 10 cognitive bias screening sessions [15]) during the 10CBS tests than did their ‘pessimistic’ conspecifics (t(11) = 3.56, p = 0.005, Fig. 3A and C). The differences in the cognitive bias index between the ‘optimists’ and ‘pessimists’ did not change significantly across the CBS (no significant test × cognitive bias interaction; F(9,99) = 0.61, NS), thereby indicating the stability of the traits (Fig. 3B). The ‘optimists’ showed an AVG cognitive bias index ranging from 0.23 to 0.60, whereas the cognitive bias index in the ‘pessimists’ ranged from 0.14 to −0.50 (Fig. 3C).
101
3.2. Effects of traits ‘optimism’ and ‘pessimism’ on rSMT performance During the rSMT training, the ‘optimists’ reached the criteria after 61 ± 1.6 training sessions, whereas the ‘pessimists’ reached the criteria after 59 ± 0.14 training sessions, respectively. We observed no significant differences in the total duration of the training between the ‘optimists’ and ‘pessimists’ (t(11) = 1.25, NS). As animals, trained previously in the ACI paradigm, knew already how to press the levers, training on the FR1 took only 2 sessions for all rats. Because statistical analysis of responding to different light patterns in one and two light trials revealed no significant cognitive judgment bias × type of light pattern interaction (F(1,11) = 0.06 and 0.02 respectively, NS) the data within twolight trials ((ON–ON–OFF) and (OFF–ON–ON)) and one-light trials ((ON–OFF–OFF) and (OFF–OFF–ON)) were pooled for further analyses. Our results indicated no statistically significant differences between the ‘optimistic’ and ‘pessimistic’ groups of animals in the proportion of pressing the ‘collect’ lever in answer to ‘clear win’, ‘near miss’ and ‘near loss’ outcomes (t(11) = 0.65, 0.34 and 0.99, respectively, NS, Fig. 4A) . Nevertheless, the ‘optimists’ were statistically more likely to press the ‘collect’ lever after the hopeless ‘clear loss’ outcome than were their ‘pessimistic’ conspecifics (t(11) = 2.39, p = 0.036, Fig. 4A). The ‘optimistic’ and ‘pessimistic’ animals did not differ in the average numbers of trials finished per session (t(11) = 0.67, NS, Fig. 4B and C). The ‘optimistic’ and ‘pessimistic’ groups of animals did not differ in the latency to press the ‘collect’ lever following ‘clear win’ (t(11) = 0.45, NS), ‘near miss’ (t(11) = 1.0 and 1.3, NS, for ON–ON–OFF and OFF–ON–ON respectively), ‘near loss’ (t(11) = 1.45 and 2.2, NS, for ON–OFF–OFF and OFF–OFF–ON respectively) and ‘clear loss’ (t(11) = 0.75, NS) outcomes. The animals’ performance was stable over the 5 test sessions. 2 way repeated measures ANOVAs with factors of test (5 levels: test 1–5) and cognitive bias index (2 levels: ‘optimism’ and ‘pessimism’) performed on both criteria (criterion 1: at least 25 trials per session, and criterion 2: <50% errors on ‘clear loss’ trials per session) revealed no significant test × cognitive bias index interaction neither in the case of criterion 1 (F(4,44) = 1.56, NS) nor criterion 2 (F(4,44) = 0.28, NS) indicating stability of performance. During 5 test sessions the ‘optimists’ completed in average 39.7 ± 2.9 trials per session, whereas the ‘pessimists’ completed in average 36.7 ± 3.2 trials per session. T-test revealed no significant differences between experimental groups (t(11) = 0.67, NS).
4. Discussion In the present study, we used an animal model to examine whether the traits of ‘optimism’ and ‘pessimism’ are associated with differences in the interpretation of outcomes of gambling in the rat slot machine task. Our results indicated that, although the ‘optimistic’ and ‘pessimistic’ animals did not differ in their propensity to interpret the ‘clear win’ (light pattern 3 × ON), near miss’ (light pattern 2 × ON) or ‘near loss (light pattern 1 × ON) outcomes as ‘win’ trials, they differed significantly in the interpretation of the ‘clear loss’ (light pattern 3 × OFF) outcomes. ‘Optimists’ were statistically more likely than were the ‘pessimists’ to press the ‘collect’ lever following the ‘clear loss’ outcome, suggesting more positive interpretation of the clearly hopeless situation. When considered with previous reports [14,15], the results of the present study demonstrate that in rats, the valence of cognitive judgment bias is an enduring behavioral trait that may determine other aspects of the animals’ behavior. As shown in Fig. 3, the
102
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
Fig. 3. Cognitive bias as a stable and enduring behavioral trait. (A) The mean ± SEM of the ‘optimism frequency’ (number of ACI tests resulting in a cognitive bias index above 0.14 out of the 10 ACI tests comprising the cognitive bias screening) of the animals that were classified as ‘pessimistic’ (filled bars, N = 7) and ‘optimistic’ (open bars, N = 6). (B) The mean ± SEM of the cognitive bias index of the animals that were classified as ‘pessimistic’ (filled circles, N = 7) and ‘optimistic’ (open circles, N = 6) across all 10 baseline ACI tests. (C) Histogram of the ‘optimism’ frequency in relation to the valence of the individual cognitive bias index (AVG from cognitive bias screening) in all (N = 13) animals. In the inset: the mean ± SEM ‘optimism’ frequency of the animals classified (based on 10 ACI tests) as ‘optimistic’ (open bar, N = 6) and ‘pessimistic’ (filled bar, N = 7).
value of this trait is both qualitative (height of cognitive bias index) and quantitative (frequency of ‘optimism’). Analysis of the lever responses during tests revealed that the animals classified as ‘optimistic’ were both more ‘optimistic’ (made a significantly higher proportion of positive lever presses in response to the ambiguous cue) and less ‘pessimistic’ (made a significantly lower proportion of negative lever presses in response to the ambiguous cue) than were their ‘pessimistic’ conspecifics (Fig. 2). This pattern was similar to a previously described pattern [14,15]. Nevertheless, the primary purpose of our research was to investigate whether ‘optimism’ and ‘pessimism’ traits could interact/influence the behavior of animals in the rSMT. Previous research in humans demonstrated that, during gambling, optimists do not benefit from their upbeat disposition [5]. In this study, relative to pessimists, optimists were more likely to develop and maintain unrealistic gambling expectations, perceive losses as near wins, and persist at gambling in the face of losses [5]. Our data supports similar phenomenon in animals. Although we did not observe significant differences between ‘optimists’ and ‘pessimists’ in responding to ‘clear wins’, ‘near misses’ or ‘near loses’, the increased propensity to ‘bet’ (press the ‘collect’ lever) after hopeless ‘clear losses’ seems to be analogous to human behavior. Indeed, in the study by Gibson and Sanbonmatsu [5] optimists, when compared to pessimists, were observed to harbor more positive gambling expectations, to maintain positive expectations, to
gamble even after negative outcomes and to not reduce their bets following losses [5]. We would like to propose several plausible explanations of this phenomenon that are convergent with those that were used to explain similar observation in humans. First, differences between ‘optimists’ and ‘pessimists’ exist in memory formation. Although, in our study, we did not investigate the differences in learning and memory between ‘optimistic’ and ‘pessimistic’ rats, it is likely that the ‘pessimistic’ animals, similar to pessimistic humans, remembered a worse outcome of their game than was their actual experience. Meanwhile, ‘optimistic’ animals tended to overestimate their past performance. In the study by Gibson and Sanbonmatsu [5] in humans, after losing, optimists recalled significantly more near wins than did pessimists. Additionally, the differences between ‘optimists’ and ‘pessimists’ in their response to ‘clear loss’ outcomes could be attributed to the differences in feedback sensitivity. The relationship between feedback sensitivity and optimism is bi-directional. Optimism decreases the impact of negative and increases the meaning of positive feedback. In humans, optimistic thoughts result in active coping strategies and weave a sense of self-efficacy and mastery over one’s environment (internal locus-of-control), which further reinforces the proactive attitude [25]. In contrast, pessimism facilitates a passive attitude, which hinders and minimizes positive and maximizes negative feedback, thereby exacerbating ‘learned
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
103
Fig. 4. The ‘optimistic‘ rats are significantly more likely to press the ‘collect’ lever after the hopeless ‘clear loss’ outcome than are their ‘pessimistic’ conspecifics in the rSMT. (A) The mean ± SEM percent of the ‘collect’ lever presses following the ‘clear win’, the ‘near miss’, the ‘near loss’ and the ‘clear loss’ outcomes in the rSMT made by animals classified (based on 10 ACI tests) as ‘pessimistic’ (filled bar, N = 7) vs. ‘optimistic’ (open bar, N = 6). * indicates significant (p ≤ 0.05) differences between the ‘optimistic’ and ‘pessimistic’ animals. Animals’ performance was stable over 5 rSMT sessions in both tested groups of rats: ‘optimistic’— panel B and ‘pessimistic’— panel C.
helplessness’ thinking patterns [25]. It is likely that the increased propensity to interpret hopeless outcomes as ‘win’ trials observed in ‘optimistic’ rats reflects a decreased sensitivity to negative feedback, thereby reinforcing the proactive attitude. Further studies using probabilistic reversal learning tests [26] are necessary to confirm this hypothesis.
Finally, the increased proportion of the ‘collect’ lever presses following the ‘clear loss’ outcome, may suggest an increased motivation to gain reward. We have recently shown that in rats, trait ‘optimism’ is associated with higher approach motivation in the progressive ratio schedule of reinforcement [15]; however, this hypothesis does not explain why the animals differed only in the hopeless situation of the ‘clear loss’ but not following other
104
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
outcomes of the rSMT. Further studies using other rodent gambling models are needed to clarify the nature of the observed differences. Although in the present study we did not analyze neurochemical correlates of investigated processes, we suggest important role for dopamine. Indeed, recent studies clearly showed that dopamine mediates optimistic bias in humans [27] and in rats [28]. Similarly, dopamine has been crucially linked with behavior of rats in the rSMT [16,18]. There are, of course, also potential confounds to the present study, which need to be discussed. First, in the ACI paradigm the tones were not counter-balanced across positively and negatively valenced outcomes. The 2 kHz tone always predicted reward and 9 kHz tone always predicted punishment. This simplification has been made basing on our previous studies [14,21,23] and unpublished observations, which clearly show that animals are not biased toward any particular tone frequency. Second potential confound could be the use of tone cues in both behavioral tasks, which due to potential generalization could affect animals’ performance. This is however unlikely, because the tones used in the rSMT and ACI tests were clearly distinct (4 kHz and 15 kHz in the rSMT vs. 2 kHz and 9 kHz in the ACI). Similarly, although the ambiguouscues were not reinforced, the probability that animals could learn about this lack of contingency is negligible. The ambiguous-cues were relatively infrequent (10 out of 50 tones presented during single ACI session) and the ACI test sessions were performed in one week intervals what minimized mentioned above possibility. Moreover, analysis of the lever presses showed lack of differences in the proportion of omitted responses to the reference and ambiguous tones (2D). Another potential confound could be a possibility that animals exposed in the ACI paradigm to electric footshocks developed learned helplessness. This is however also unlikely as learned helplessness is a state in which an organism forced to bear aversive stimuli, becomes unable or unwilling to avoid subsequent encounters with those stimuli, even if they are ‘escapable’, because it has learned that it cannot control the situation. That was certainly not the case in our experimental design. During the training the animals learned to control the situation and to avoid unpleasant shock by pressing the ‘negative’ lever in answer to the tone predicting incoming punishment. As shown in Fig. 2C, the animals taking part in the experiment avoid successfully about 80% of shocks and remaining 20% are terminated after their onset. Moreover, because there were no significant differences between ‘optimists’ and ‘pessimists’ neither in the length of the negative tone training nor in the proportion of negative lever presses in response to the negative tone, it is unlikely that ‘pessimistic’ animals are more sensitive to footshocks or more likely to develop learned helplessness than their ‘optimistic’ counterparts. Finally, optimistic animals made more responses to ‘positive’ lever not only in response to the ambiguous-cues but also following the presentation of ‘positive’ reference tones, what might indicate a general increase in anticipation of reward. Although ideally, the differences in cognitive judgment bias should be indicated only by differences in responding to the ambiguous-cues, practice shows that often the bias is generalized and manifested also by altered responding to the positive and/or negative tones. Similar effects have been observed in our previous studies [14,15], and in the landmark study by Enkel et al. [10]. Last but not least, the comorbidity of gambling disorder with depression, at first sight, may seem problematic for interpretation of the present results. However, in our study we investigated interrelation between cognitive judgment bias and propensity to gamble in the healthy subjects. The ‘optimistic’ and ‘pessimistic’ judgment biases described in our study did not reflect pathological, depressive or maniacal states but were just behavioral traits of individual animals and as such could determine propensity of rats to certain aspects of gambling. Further studies should investigate
the co-morbidity of depression and gambling disorder in an animal models of depression e.g. based on chronic stress. 5. Conclusion In our recent studies, we have demonstrated a relationship between cognitive judgment bias and the vulnerability of animals to stress-induced anhedonia [14] and motivation [15]. The results of our present study revealed that, similar to humans, the cognitive judgment bias of rats could determine their propensity to gamble. Our research provides a unique platform for modeling interactions between behavioral traits and decision-making that indicate a risk for pathological behaviors in animals. Further studies should demonstrate whether a correlation within a (relatively small) healthy population could be extrapolated to a clinical sample. Conflict of interests The authors declare that there were no conflicts of interest in this study. Acknowledgments This work was supported by the Polish National Science Center (Research grant DEC-2014/13/B/NZ4/00214 to RR) and the statutory funds of the Institute of Pharmacology Polish Academy of Sciences. We thank Marcin Rzycki for graphical support. References [1] A. Blaszczynski, D. Silove, Cognitive and behavioral therapies for pathological gambling, J. Gambl. Stud. 11 (1995) 195–220. [2] T. Gilovich, Biased evaluation and persistence in gambling, J. Pers. Soc. Psychol. 44 (1983) 1110–1126. [3] R. Ladouceur, Gambling the hidden addiction, Can. J. Psychiatry 49 (2004) 501–503. [4] C.S. Carver, M.F. Scheier, Dispositional optimism, Trends Cogn. Sci. 18 (2014) 293–299. [5] B. Gibson, D.M. Sanbonmatsu, Optimism, pessimism, and gambling: the downside of optimism, Pers. Soc. Psychol. Bull. 30 (2004) 149–160. [6] M. Bateson, S. Desire, S.E. Gartside, G.A. Wright, Agitated honeybees exhibit pessimistic cognitive biases, Curr. Biol. 21 (2011) 1070–1073. [7] E.J. Bethell, A. Holmes, A. Maclarnon, S. Semple, Evidence that emotion mediates social attention in Rhesus macaques, PLoS One 7 (2007) e44387. [8] B.O. Brilot, L. Asher, M. Bateson, Stereotyping starlings are more pessimistic, Anim. Cogn. 13 (2010) 721–731. [9] R.E. Doyle, C. Lee, V. Deiss, A.D. Fisher, G.N. Hinch, A. Boissy, Measuring judgement bias and emotional reactivity in sheep following long-term exposure to unpredictable and aversive events, Physiol. Behav. 102 (2011) 503–510. [10] T. Enkel, D. Gholizadeh, O. von Bohlen Und Halbach, C. Sanchis-Segura, R. Hurlemann, R. Spanagel, et al., Ambiguous-cue interpretation is biased under stress- and depression-like states in rats, Neuropsychopharmacology 35 (2010) 1008–1015. [11] E.J. Harding, E.S. Paul, M. Mendl, Animal behaviour: cognitive bias and affective state, Nature 427 (2004) 312. [12] M. Mendl, J. Brooks, C. Basse, O. Burman, E. Paul, E. Blackwell, et al., Dogs showing separation-related behaviour exhibit a ‘pessimistic’ cognitive bias, Curr. Biol. 20 (2010) R839–R840. [13] R. Rygula, H. Pluta, P. Popik, Laughing rats are optimistic, PLoS One 7 (2012) e51959. [14] R. Rygula, J. Papciak, P. Popik, Trait pessimism predicts vulnerability to stress-induced anhedonia in rats, Neuropsychopharmacology 38 (2013) 2188–2196. [15] R. Rygula, J. Golebiowska, J. Kregiel, J. Kubik, P. Popik, Effects of optimism on motivation in rats, Front. Behav. Neurosci. 9 (2015) 32. [16] C.A. Winstanley, P.J. Cocker, R.D. Rogers, Dopamine modulates reward expectancy during performance of a slot machine task in rats: evidence for a ‘near-miss’ effect, Neuropsychopharmacology 36 (2011) 913–925. [17] C.A. Winstanley, Gambling rats: insight into impulsive and addictive behavior, Neuropsychopharmacology 36 (2011) 359. [18] P.J. Cocker, B. Le Foll, R.D. Rogers, C.A. Winstanley, A selective role for dopamine D(4) receptors in modulating reward expectancy in a rodent slot machine task, Biol. Psychiatry 75 (2014) 817–824.
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105 [19] J. Papciak, P. Popik, E. Fuchs, R. Rygula, Chronic psychosocial stress makes rats more ‘pessimistic’ in the ambiguous-cue interpretation paradigm, Behav. Brain Res. 256 (2013) 305–310. [20] R. Rygula, J. Papciak, P. Popik, The effects of acute pharmacological stimulation of the 5-HT, NA and DA systems on the cognitive judgement bias of rats in the ambiguous-cue interpretation paradigm, Eur. Neuropsychopharmacol. 24 (2014) 1103–1111. [21] R. Rygula, E. Szczech, J. Papciak, A. Nikiforuk, P. Popik, The effects of cocaine and mazindol on the cognitive judgement bias of rats in the ambiguous-cue interpretation paradigm, Behav. Brain Res. 270 (2014) 206–212. [22] R. Rygula, J. Golebiowska, J. Kregiel, M. Holuj, P. Popik, Acute administration of lithium, but not valproate, modulates cognitive judgment bias in rats, Psychopharmacology (Berl.) 232 (2014) 2149–2156. [23] R. Rygula, E. Szczech, J. Kregiel, J. Golebiowska, J. Kubik, P. Popik, Cognitive judgment bias in the psychostimulant-induced model of mania in rats, Psychopharmacology (Berl.) 232 (2015) 651–660.
105
[24] D.C. Howell, Statistical Methods for Psychology, 4th ed., Wadsworth, Belmont, California, 1997. [25] D. Hecht, The neural basis of optimism and pessimism, Exp. Neurobiol. 22 (2013) 173–199. [26] A. Bari, D.E. Theobald, D. Caprioli, A.C. Mar, A. Aidoo-Micah, J.W. Dalley, et al., Serotonin modulates sensitivity to reward and negative feedback in a probabilistic reversal learning task in rats, Neuropsychopharmacology 35 (2010) 1290–1301. [27] T. Sharot, M. Guitart-Masip, C.W. Korn, R. Chowdhury, R.J. Dolan, How dopamine enhances an optimism bias in humans, Curr. Biol. 22 (2012) 1477–1481. [28] J. Kregiel, J. Golebiowska, P. Popik, R. Rygula, Dopamine induces an optimism bias in rats—pharmacological proof for the translational validity of the ambiguous-cue interpretation test, Behav. Brain Res. 297 (2015) 84–90.
Contents lists available at ScienceDirect
Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr
Research report
Effects of optimism on gambling in the rat slot machine task Dominik Rafa 1 , Jakub Kregiel 1 , Piotr Popik, Rafal Rygula ∗ Affective Cognitive Neuroscience Lab, Department of Behavioral Neuroscience and Drug Development, Institute of Pharmacology Polish Academy of Sciences, 12 Smetna Street, 31-343 Krakow, Poland
h i g h l i g h t s • • • •
We identified rats displaying ‘pessimistic’ and ‘optimistic’ traits. We determined how these traits affect behavior in the rat Slot Machine Task. ‘Optimistic’ rats were more likely to gamble in the hopeless ‘clear loss’ situation. We demonstrated interrelation between optimism and gambling in an animal model.
a r t i c l e
i n f o
Article history: Received 21 October 2015 Received in revised form 30 November 2015 Accepted 11 December 2015 Available online 14 December 2015 Keywords: Rat Gambling Ambiguous-cue interpretation Pessimism Optimism Cognitive judgment bias
a b s t r a c t Although gambling disorder is a serious social problem in modern societies, information about the behavioral traits that could determine vulnerability to this psychopathology is still scarce. In this study, we used a recently developed ambiguous-cue interpretation (ACI) paradigm to investigate whether ‘optimism’ and ‘pessimism’ as behavioral traits may determine the gambling-like behavior of rodents. In a series of ACI tests (cognitive bias screening), we identified rats that displayed ‘pessimistic’ and ‘optimistic’ traits. Subsequently, using the rat slot machine task (rSMT), we investigated if the ‘optimistic’/‘pessimistic’ traits could determine the crucial feature of gambling-like behavior that has been investigated in rats and humans: the interpretation of ‘near-miss’ outcomes as a positive (i.e., win) situation. We found that ‘optimists’ did not interpret ‘near-miss’, ‘near loss’, or ‘clear win’ as win trials more often than their ‘pessimistic’ conspecifics; however, the ‘optimists’ were statistically more likely to reach for a reward in the hopeless ‘clear loss’ situation. This agrees with human studies and provides a platform for modeling interactions between behavioral traits and gambling in animals. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Cognitive theories of psychiatric disorders have articulated the mediating role of cognitive biases, distortions in reasoning, and errors in judgment in the phenomenology of behavioral addictions, such as gambling disorder [1–3]. One of the errors in reasoning that may be critically involved in gambling disorder is cognitive judgment bias that is manifested by excessive optimism. According to Carver and Scheier [4], the generalized positive outcome expectancies of optimists result persistence in attempting to accomplish goals in the face of adversity, whereas the generalized negative outcome expectancies of pessimists result in withdrawal. Thus, the benefits of optimism may arise primarily in situations in which this persistence is rewarded. Unfortunately, not all situations and
∗ Corresponding author. Fax: +48 12 6374500. E-mail address: [email protected] (R. Rygula). 1 These authors contributed equally to this work. http://dx.doi.org/10.1016/j.bbr.2015.12.013 0166-4328/© 2015 Elsevier B.V. All rights reserved.
tasks have positive outcomes. Gambling is one important domain in which persistence is unlikely to be consistently rewarded and in which optimism may be a liability [5]. In agreement with this assumption, recent research on optimism and gambling has found that optimists had more positive expectations for gambling than did pessimists and were less likely to reduce their betting after poor outcomes [5]. Despite these findings, there is still a dearth of information concerning the interaction of optimism and gambling in both humans and rodents, and despite the important role of cognitive biases in psychopathologies, they have not been the subject of much preclinical behavioral research. Although several research groups over the past decade have attempted to measure the effects of various behavioral and pharmacological manipulations on ‘optimistic’ and ‘pessimistic’ judgment biases [6–13], almost none of them have investigated ‘optimism’/‘pessimism’ as an enduring and stable behavioral trait that could be used to evaluate its implications for psychiatric conditions, such as gambling disorder.
98
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
We have recently shown that, similar to humans, the cognitive judgment bias of animals has components of both enduring traits and transient states [14]. A trait is the stable level of an individual’s ‘pessimism’/‘optimism’ that is generally experienced, whereas a state represents the valence of cognitive judgment bias that may change based upon the situation or contextual factors. We also have identified a relationship between cognitive judgment bias as a behavioral trait and vulnerability of animals to stress-induced anhedonia [14] as well as a correlation between the level of optimism and motivation [15]. In the present study, we used multiple ambiguous-cue interpretation (ACI) tests [10,11,13] to isolate two groups of rats that consistently differed in their cognitive judgment bias over time [14,15]. These two groups of ‘pessimistic’ and ‘optimistic’ rats were subsequently trained and tested for the propensity to gamble in the rat slot machine task (rSMT) [16]. In this behavioral paradigm subjects respond to a series of three flashing lights, loosely analogous to the wheels of a slot machine, causing the lights to set to ‘ON’ or ‘OFF’. A winning outcome is signaled if all three lights were illuminated. At the end of each trial, rats chose between responding on the ‘collect’ lever, resulting in reward on win trials, but a time penalty on loss trials, or starting a new trial. Recent studies using this task revealed important role of dopamine in pathological gambling [16–18]. We hypothesized that, similar to humans, rats display different propensities to gamble in association with traits of ‘optimism’ and ‘pessimism.’
2. Methods 2.1. Ethics statement All described experimental procedures were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Committee for Ethics in Animal Experiments at the Institute of Pharmacology Polish Academy of Sciences. 2.2. Subjects and housing We used 24 male Sprague Dawley rats (Charles River, Germany) that weighed between 175 and 200 g upon arrival. The animals were housed in groups of five, in a temperature (21 ± 1 ◦ C) and humidity (40–50%) controlled room under a 12/12 h dark/light cycle (lights on at 07:00 h). The animals were habituated to the housing conditions and experimental facility for two weeks after arrival and before the start of experiments. In all experiments, the animals received 15–20 g of food per rat per day (standard laboratory chow), which corresponds to mild food-deprivation. The food deprivation began seven days prior to beginning of the training. The rats were weighed once a week. The water was provided ad libitum . The animals were trained and tested during the light phase of the dark/light cycle. The rats were habituated to the experimental room for 30 min prior to training and testing sessions and were tested once daily. 2.3. Ambiguous-cue interpretation test 2.3.1. Apparatus The experiments were conducted in eight computer-controlled, operant conditioning boxes (Med Associates, St. Albans, Vermont, USA). The boxes were equipped with lights, speakers, liquid dispensers (0.1 ml of 5% sucrose solution), electric grid floors, and two retractable levers. The levers were located on both sides of the
liquid dispenser. The experimental protocols were written in Med State notation code (Med Associates). 2.3.2. Behavioral training The experimental training and testing procedures for the ACI paradigm that was used in this study were modified from procedures previously described by Enkel et al. [10] and have been described in detail elsewhere [13,14,19–23]. In brief, initially the animals were trained to press one lever when a ‘positive’ tone (2000 Hz at 75 dB) signaled the availability of a reward (5% sucrose solution) and to press second lever when another, ‘negative’ tone (9000 Hz at 75 dB) signaled a forthcoming punishment (0.5 mA, 10 s). By pressing appropriate levers the animals could either receive a reward or avoid punishment. The tone presentations were separated by 10 s intertrial intervals (ITI) and each training session lasted 30 min. The animals had to fulfill the criteria of at least 90% of accurate responses to the tone signaling reward availability maintained over three consecutive training sessions and, at least 60% of correct punishment-prevention responses maintained over three consecutive training sessions, to proceed to the discrimination training. During the discrimination-training phase, the animals were trained to discriminate between pseudo-randomly presented positive (20) and negative (20) tones by responding to the appropriate levers (as learned in the previous training stages), thereby minimizing punishment and maximizing reward delivery. Each discrimination training session lasted 40 min. The animals had to achieve a minimum of 70% correct responses with each lever, maintained over three consecutive discrimination sessions to be qualified for the ACI testing. As for an animal it is really difficult to learn instrumental avoidance reaction in anticipation of punishment, the criterion at the stage of ‘negative tone’ training was lower than for the ‘positive tone’ training. However, at the stage of discrimination, all animals were trained until they reached equally stable discrimination ratio for both tones—minimum 70%. 2.3.3. Ambiguous-cue testing During the ACI testing sessions the animals were exposed to 20 negative, 20 positive, and 10 ambiguous (5000 Hz at 75 dB) tone presentations. The tones were played in a pseudo-randomized order and were separated by 10 s ITIs. The responses to each tone (positive, ambiguous and negative) during the ACI testing were analyzed as the proportion of the overall number of responses to a given tone. To calculate the cognitive bias index, we subtracted the proportion of negative responses to the ambiguous-cues from the proportion of positive responses to the ambiguous-cue, which resulted in values ranging between −1 and 1. After median split, values above 0.14 indicated an overall positive judgment and ‘optimistic’ interpretation of the ambiguous-cue while the values below 0.14 indicated overall negative judgment and ‘pessimism’. 2.3.4. Cognitive bias screening (CBS) The CBS procedure has been described in detail elsewhere [14]. Briefly, to assess the cognitive judgment bias as a trait, we examined the animals using a series of 10 consecutive ACI tests that were conducted at one-week intervals. Based upon the average cognitive bias index that was obtained from these 10 ACI tests, the rats were divided into two subgroups: ‘optimistic’ and ‘pessimistic’. 2.4. Rat slot machine task test The experimental set-up, training and testing procedures for the rSMT paradigm used in this study were adapted and modified from procedures that were previously described by Winstanley et al. [16].
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
99
Following the CBS screening, the animals were re-trained for the rSMT. To minimize the impact of one task on the other, the animals were re-trained and tested by another experimenter in another experimental room, using different set of operant boxes (Coulbourn Instruments, Whitehall, PA, USA). 2.4.1. Apparatus The experiments were conducted in four computer-controlled, operant conditioning boxes (Coulbourn Instruments, USA) each enclosed within a ventilated sound-attenuating cabinet. Each box was fitted with the loudspeaker, house light, two retractable levers and an array of three nose poke response holes that were positioned 2 cm above a bar floor on one of the walls. A stimulus light was set at the back of each hole. Infrared beams detected nose-poke responses into these apertures. A food tray was located in the middle of the opposite wall with one retractable lever located on each side. One of the levers served as the ‘collect’ lever. By pressing it, the rat indicated a choice that suggested that the animal assumed the trial was the ‘win’ trial. The other retractable lever was a ‘roll’ lever. By pressing it, the rat indicated that the animal chose not to collect the reward but rather to initiate or continue the game. The loudspeaker was located above the food tray. The apparatuses were controlled by Graphic State 3.0 software (Coulbourn Instruments, USA). 2.4.2. Behavioral training All training and testing stages were adapted from Winstanley et al. [16]. In brief, initially the animals were placed in the operant chamber with a house light turned on and with sucrose pellets (45 mg, Bio-Serv, New Jersey, USA) located in the food tray and nose-poke apertures. Each session lasted 30 min. Then, rats were trained to press the levers to earn the food reward under a fixed ratio 1 (FR1) schedule of reinforcement. Only one lever was presented during the session. When the animal had made more than 50 lever presses in a session, training was repeated on the other lever. The order in which the levers were presented (left/right) was counterbalanced between the subjects. Once the animals had learnt to press the levers, the proper training for the rSMT began. 2.4.3. Rat slot machine task Following the completion of successive versions of the slot machine program, which gradually increased in complexity (described in detail by Winstanley et al. [16] and Cocker et al. [18]), the proper rSMT training and testing began. The rat had to press the ‘roll’ lever that triggered its retraction and started the trial by flashing the first hole light at a frequency of 2 Hz (Fig. 1). When the rat nose-poked into the first hole, depending on the schedule, the light inside the hole was either turned ON or turned OFF for the rest of the trial. The ON or OFF status of the hole was accompanied by either a high (15 000 Hz at 68 dB) or low (4000 Hz at 68 dB) acoustic tone, which served as an additional cue. The sounds lasted for 1 s. Following the first hole step, the light in second hole began to flash. When the rat made a nose-poke response into the second hole, as in the previous step, the light inside was either turned ON or turned OFF, depending on the schedule, and a high or low relevant acoustic tone was presented for 1 s. Then, the light inside the third hole began to flash, and the sequence of events (depending on the schedule, light ON/OFF accompanied with a relevant acoustic signal) was repeated, similar to the procedure for the first and second holes. At this moment, the animal was exposed to and, by nose-poking, responded to the series of three stimuli. To increase the number of meaningful trials, we modified the original procedure that was described by Winstanley et al. [16], and reduced the number of possible two and three light pattern outcomes to (ON–ON–OFF or OFF–ON–ON) and (ON–OFF–OFF and OFF–OFF–ON), respectively. In this way each trial could have
Fig. 1. (A) Schematic representation of the experimental schedule and main result of the study. (B) Scheme of the rSMT test session.
only six out of eight possible patterns: ON–ON–ON, ON–ON–OFF, OFF–ON–ON, ON–OFF–OFF, OFF–OFF–ON and OFF–OFF–OFF. This gave one state with three lights ON (a rewarded, clear ‘win’), two near-miss states with two lights ON (ON–ON–OFF or OFF–ON–ON, ‘near miss’), two states with one light ON (ON–OFF–OFF, OFF–OFF–ON, ‘near loss’) and one state with all lights OFF (a clear ‘loss’).
100
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
For analysis, the data from two-light trials ((ON–ON–OFF) and (OFF–ON–ON)) and one-light trials ((ON–OFF–OFF) and (OFF–OFF–ON)) were pooled. The frequency of all six light patterns was predefined in a pseudo-random order and each pattern was presented once per block of six trials so that the animals could obtain a food reward only once in each block. The pseudo-random order of trials in each block was different so that animals could not learn the sequence. The rat was required to interpret the given pattern as a ‘win’ or ‘loss’. This was accomplished at the ‘decision point’ by extending both the ‘roll’ and ‘collect’ levers. On the clear ‘win’ trials, during which all three lights were set to ON, pressing the ‘collect’ lever resulted in a delivery of five sucrose pellets. On any other pattern, with at least one light being turned OFF, responding on the ‘collect’ lever resulted in a penalty of 25 s of time-out, during which the animal could not quickly start a new trial. However, pressing the ‘roll’ lever on any trial cancelled the potential reward or the time-out, so that the rat could immediately start a new trial. The rat’s optimal strategy was to respond with the ‘collect’ lever only on ‘win’ trials to obtain the food reward and to respond with a ‘roll’ lever on every ‘loss’ trial to avoid the time-out and start a new trial. There was no time requirement to respond on the lever or the nose-poke. Rats were trained once a day, five days per week, until they established a stable performance. The animals were deemed to have successfully acquired the task if they completed ≥25 trials per session and made ≤50% collect responses on clear loss (OFF–OFF–OFF) trials, five times in a row. 2.5. Experimental design and behavioral measures The experimental design is schematically presented in Fig. 1. After establishing the ‘optimistic’ and ‘pessimistic’ traits in the individual animals (CBS procedure), the rats were divided into two experimental groups: ‘optimistic’ and ‘pessimistic’ (Fig. 1A). To assess whether the traits of ‘optimism’ and ‘pessimism’ may determine the propensity of the animals to gamble, the rats were re-trained in another set of four experimental boxes (Coulbourn Instruments, USA) and subsequently tested for five consecutive days on the rSMT. After reaching the training criteria (described above), the animals were tested five times on the rSMT. Average results from these five tests were compared between the groups of animals that were previously classified as ‘optimistic’ and ‘pessimistic’. 2.6. Statistics We analyzed the data using SPSS (version 21.0, SPSS Inc., Chicago, IL, United States). The differences in the processing of the experimental tones between the ‘optimists’ and ‘pessimists’ were investigated using a three-way analysis of variance (ANOVA) with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factors of lever (two levels: positive and negative) and tone (three levels: positive, ambiguous and negative). The stability of the trait was analyzed using a two-way ANOVA with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factor of test (10 levels: 1–10). The effects of different light patterns within 2 and 1-light trials on responding to ‘collect’ lever were analyzed using a two-way ANOVA with the between-subject factor of cognitive judgment bias (two levels: ‘optimistic’ and ‘pessimistic’) and the within-subjects factor of the type of light pattern (2 levels: ON–ON–OFF and OFF–ON–ON for 2 light trials, and 2 levels: ON–OFF–OFF and OFF–OFF–ON for 1 light trials). The differences between ‘optimistic’ and ‘pessimistic’ animals in the length of training, ‘optimism’ frequency, numbers of trials finished per experimental session, responding to the ‘collect’ lever at different light patterns in the rSMT and response latencies
Fig. 2. ‘Optimistic’ vs. ‘pessimistic’ animals; results of the cognitive bias screening. (A) The mean ± SEM of the cognitive bias index of the animals that were classified (based on 10 ACI tests) as ‘pessimistic’ (circles, N = 7) vs. ‘optimistic’ (squares, N = 6). A cognitive bias index above 0.14 indicates an overall positive judgment and ‘optimistic’ interpretation of the ambiguous cue. (B) The mean ± SEM of the proportion of positive, (C) negative and (D) omitted responses to the trained and ambiguous tones in the ‘optimistic’ (open circles, N = 6) and ‘pessimistic’ (filled circles, N = 7) rat groups. * indicates significant (p ≤ 0.05) differences between the ‘optimistic’ and ‘pessimistic’ animals.
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
were analyzed separately using t-tests. For pair-wise comparisons, we adjusted the values using Sidak’s correction factor for multiple comparisons [24]. All tests of significance were performed at ˛ = 0.05. We tested the homogeneity of variance using Levene’s test and, for repeated-measures analyses, we confirmed the sphericity using Mauchly’s test. The data are presented as the mean ± SEM.
3. Results Out of the 24 initially trained animals, 24 qualified for CBS using the ACI test; however, after re-training in the rSMT, only 13 reached the training criteria and qualified for the final testing. It is worth noticing that the animals were excluded, because they were unable to successfully acquire the rSMT task by completion of more than 25 trials per session and not because they could not make ≤50% collect responses on clear loss (OFF–OFF–OFF) trials, five times in a row. Out of 11 excluded animals 6 were ‘optimistic’ and 5 were ‘pessimistic’. During the ACI training, the ‘optimists’ reached the criteria of ‘positive’ tone, ‘negative’ tone and discrimination trainings after 13.2 ± 0.1, 30.0 ± 3.7 and 43.2 ± 1.2 training sessions, respectively, whereas the ‘pessimists’ reached the criteria after 13.1 ± 0.14, 27.1 ± 4.2 and 42.1 ± 1.0 training sessions, respectively. We observed no significant differences in the length of ‘positive’ tone training (t(11) = 0.11, NS), ‘negative’ tone training (t(11) = 0.50, NS), discrimination training (t(11) = 0.79, NS) and in the total duration of training (t(11) = 0.68, NS) between the ‘optimists’ (86.4 ± 4.4 training sessions) and ‘pessimists’ (82.3 ± 3.8 training sessions). ‘Optimists’ and ‘pessimists’ did not significantly differ in body weight (t(11) = 0.13, NS).
3.1. ‘Pessimistic’ vs. ‘optimistic’ rats Based on the results of the CBS (Fig. 3A), we divided the animals that qualified for the rSMT into two groups using the median split. These groups were clearly distinctive in their interpretation of the ambiguous-cues over time: ‘optimists’ (N = 6, average (AVG) cognitive bias index >0.14) and ‘pessimists’ (N = 7, AVG cognitive bias index <0.14). Further analysis revealed significant differences in the patterns of lever responses between ‘optimists’ and ‘pessimists’ (lever × tone × cognitive bias interaction (F(2,22) = 11.4, p < 0.001)). The ‘optimists’ responded significantly more often to the positive lever in response to the ambiguous and negative tones tone when compared to ‘pessimists’ (p < 0.002 and p = 0.033, respectively, Fig. 2B) and less often to the negative lever in response to the positive and negative tones (p = 0.002 and p = 0.005, respectively, Fig. 2C). As shown in Fig. 2D, ‘optimists’ and ‘pessimists’ did not differ in the numbers of omitted trials (no significant effects of cognitive bias (F(1,11) = 0.002, NS) or in the cognitive bias × tone interaction (F(2,22) = 1.87, NS)). ‘Optimistic’ animals showed significantly higher average frequencies of ‘optimism’ (number of tests when the cognitive bias index of an individual animal was higher than zero, out of the 10 cognitive bias screening sessions [15]) during the 10CBS tests than did their ‘pessimistic’ conspecifics (t(11) = 3.56, p = 0.005, Fig. 3A and C). The differences in the cognitive bias index between the ‘optimists’ and ‘pessimists’ did not change significantly across the CBS (no significant test × cognitive bias interaction; F(9,99) = 0.61, NS), thereby indicating the stability of the traits (Fig. 3B). The ‘optimists’ showed an AVG cognitive bias index ranging from 0.23 to 0.60, whereas the cognitive bias index in the ‘pessimists’ ranged from 0.14 to −0.50 (Fig. 3C).
101
3.2. Effects of traits ‘optimism’ and ‘pessimism’ on rSMT performance During the rSMT training, the ‘optimists’ reached the criteria after 61 ± 1.6 training sessions, whereas the ‘pessimists’ reached the criteria after 59 ± 0.14 training sessions, respectively. We observed no significant differences in the total duration of the training between the ‘optimists’ and ‘pessimists’ (t(11) = 1.25, NS). As animals, trained previously in the ACI paradigm, knew already how to press the levers, training on the FR1 took only 2 sessions for all rats. Because statistical analysis of responding to different light patterns in one and two light trials revealed no significant cognitive judgment bias × type of light pattern interaction (F(1,11) = 0.06 and 0.02 respectively, NS) the data within twolight trials ((ON–ON–OFF) and (OFF–ON–ON)) and one-light trials ((ON–OFF–OFF) and (OFF–OFF–ON)) were pooled for further analyses. Our results indicated no statistically significant differences between the ‘optimistic’ and ‘pessimistic’ groups of animals in the proportion of pressing the ‘collect’ lever in answer to ‘clear win’, ‘near miss’ and ‘near loss’ outcomes (t(11) = 0.65, 0.34 and 0.99, respectively, NS, Fig. 4A) . Nevertheless, the ‘optimists’ were statistically more likely to press the ‘collect’ lever after the hopeless ‘clear loss’ outcome than were their ‘pessimistic’ conspecifics (t(11) = 2.39, p = 0.036, Fig. 4A). The ‘optimistic’ and ‘pessimistic’ animals did not differ in the average numbers of trials finished per session (t(11) = 0.67, NS, Fig. 4B and C). The ‘optimistic’ and ‘pessimistic’ groups of animals did not differ in the latency to press the ‘collect’ lever following ‘clear win’ (t(11) = 0.45, NS), ‘near miss’ (t(11) = 1.0 and 1.3, NS, for ON–ON–OFF and OFF–ON–ON respectively), ‘near loss’ (t(11) = 1.45 and 2.2, NS, for ON–OFF–OFF and OFF–OFF–ON respectively) and ‘clear loss’ (t(11) = 0.75, NS) outcomes. The animals’ performance was stable over the 5 test sessions. 2 way repeated measures ANOVAs with factors of test (5 levels: test 1–5) and cognitive bias index (2 levels: ‘optimism’ and ‘pessimism’) performed on both criteria (criterion 1: at least 25 trials per session, and criterion 2: <50% errors on ‘clear loss’ trials per session) revealed no significant test × cognitive bias index interaction neither in the case of criterion 1 (F(4,44) = 1.56, NS) nor criterion 2 (F(4,44) = 0.28, NS) indicating stability of performance. During 5 test sessions the ‘optimists’ completed in average 39.7 ± 2.9 trials per session, whereas the ‘pessimists’ completed in average 36.7 ± 3.2 trials per session. T-test revealed no significant differences between experimental groups (t(11) = 0.67, NS).
4. Discussion In the present study, we used an animal model to examine whether the traits of ‘optimism’ and ‘pessimism’ are associated with differences in the interpretation of outcomes of gambling in the rat slot machine task. Our results indicated that, although the ‘optimistic’ and ‘pessimistic’ animals did not differ in their propensity to interpret the ‘clear win’ (light pattern 3 × ON), near miss’ (light pattern 2 × ON) or ‘near loss (light pattern 1 × ON) outcomes as ‘win’ trials, they differed significantly in the interpretation of the ‘clear loss’ (light pattern 3 × OFF) outcomes. ‘Optimists’ were statistically more likely than were the ‘pessimists’ to press the ‘collect’ lever following the ‘clear loss’ outcome, suggesting more positive interpretation of the clearly hopeless situation. When considered with previous reports [14,15], the results of the present study demonstrate that in rats, the valence of cognitive judgment bias is an enduring behavioral trait that may determine other aspects of the animals’ behavior. As shown in Fig. 3, the
102
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
Fig. 3. Cognitive bias as a stable and enduring behavioral trait. (A) The mean ± SEM of the ‘optimism frequency’ (number of ACI tests resulting in a cognitive bias index above 0.14 out of the 10 ACI tests comprising the cognitive bias screening) of the animals that were classified as ‘pessimistic’ (filled bars, N = 7) and ‘optimistic’ (open bars, N = 6). (B) The mean ± SEM of the cognitive bias index of the animals that were classified as ‘pessimistic’ (filled circles, N = 7) and ‘optimistic’ (open circles, N = 6) across all 10 baseline ACI tests. (C) Histogram of the ‘optimism’ frequency in relation to the valence of the individual cognitive bias index (AVG from cognitive bias screening) in all (N = 13) animals. In the inset: the mean ± SEM ‘optimism’ frequency of the animals classified (based on 10 ACI tests) as ‘optimistic’ (open bar, N = 6) and ‘pessimistic’ (filled bar, N = 7).
value of this trait is both qualitative (height of cognitive bias index) and quantitative (frequency of ‘optimism’). Analysis of the lever responses during tests revealed that the animals classified as ‘optimistic’ were both more ‘optimistic’ (made a significantly higher proportion of positive lever presses in response to the ambiguous cue) and less ‘pessimistic’ (made a significantly lower proportion of negative lever presses in response to the ambiguous cue) than were their ‘pessimistic’ conspecifics (Fig. 2). This pattern was similar to a previously described pattern [14,15]. Nevertheless, the primary purpose of our research was to investigate whether ‘optimism’ and ‘pessimism’ traits could interact/influence the behavior of animals in the rSMT. Previous research in humans demonstrated that, during gambling, optimists do not benefit from their upbeat disposition [5]. In this study, relative to pessimists, optimists were more likely to develop and maintain unrealistic gambling expectations, perceive losses as near wins, and persist at gambling in the face of losses [5]. Our data supports similar phenomenon in animals. Although we did not observe significant differences between ‘optimists’ and ‘pessimists’ in responding to ‘clear wins’, ‘near misses’ or ‘near loses’, the increased propensity to ‘bet’ (press the ‘collect’ lever) after hopeless ‘clear losses’ seems to be analogous to human behavior. Indeed, in the study by Gibson and Sanbonmatsu [5] optimists, when compared to pessimists, were observed to harbor more positive gambling expectations, to maintain positive expectations, to
gamble even after negative outcomes and to not reduce their bets following losses [5]. We would like to propose several plausible explanations of this phenomenon that are convergent with those that were used to explain similar observation in humans. First, differences between ‘optimists’ and ‘pessimists’ exist in memory formation. Although, in our study, we did not investigate the differences in learning and memory between ‘optimistic’ and ‘pessimistic’ rats, it is likely that the ‘pessimistic’ animals, similar to pessimistic humans, remembered a worse outcome of their game than was their actual experience. Meanwhile, ‘optimistic’ animals tended to overestimate their past performance. In the study by Gibson and Sanbonmatsu [5] in humans, after losing, optimists recalled significantly more near wins than did pessimists. Additionally, the differences between ‘optimists’ and ‘pessimists’ in their response to ‘clear loss’ outcomes could be attributed to the differences in feedback sensitivity. The relationship between feedback sensitivity and optimism is bi-directional. Optimism decreases the impact of negative and increases the meaning of positive feedback. In humans, optimistic thoughts result in active coping strategies and weave a sense of self-efficacy and mastery over one’s environment (internal locus-of-control), which further reinforces the proactive attitude [25]. In contrast, pessimism facilitates a passive attitude, which hinders and minimizes positive and maximizes negative feedback, thereby exacerbating ‘learned
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
103
Fig. 4. The ‘optimistic‘ rats are significantly more likely to press the ‘collect’ lever after the hopeless ‘clear loss’ outcome than are their ‘pessimistic’ conspecifics in the rSMT. (A) The mean ± SEM percent of the ‘collect’ lever presses following the ‘clear win’, the ‘near miss’, the ‘near loss’ and the ‘clear loss’ outcomes in the rSMT made by animals classified (based on 10 ACI tests) as ‘pessimistic’ (filled bar, N = 7) vs. ‘optimistic’ (open bar, N = 6). * indicates significant (p ≤ 0.05) differences between the ‘optimistic’ and ‘pessimistic’ animals. Animals’ performance was stable over 5 rSMT sessions in both tested groups of rats: ‘optimistic’— panel B and ‘pessimistic’— panel C.
helplessness’ thinking patterns [25]. It is likely that the increased propensity to interpret hopeless outcomes as ‘win’ trials observed in ‘optimistic’ rats reflects a decreased sensitivity to negative feedback, thereby reinforcing the proactive attitude. Further studies using probabilistic reversal learning tests [26] are necessary to confirm this hypothesis.
Finally, the increased proportion of the ‘collect’ lever presses following the ‘clear loss’ outcome, may suggest an increased motivation to gain reward. We have recently shown that in rats, trait ‘optimism’ is associated with higher approach motivation in the progressive ratio schedule of reinforcement [15]; however, this hypothesis does not explain why the animals differed only in the hopeless situation of the ‘clear loss’ but not following other
104
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105
outcomes of the rSMT. Further studies using other rodent gambling models are needed to clarify the nature of the observed differences. Although in the present study we did not analyze neurochemical correlates of investigated processes, we suggest important role for dopamine. Indeed, recent studies clearly showed that dopamine mediates optimistic bias in humans [27] and in rats [28]. Similarly, dopamine has been crucially linked with behavior of rats in the rSMT [16,18]. There are, of course, also potential confounds to the present study, which need to be discussed. First, in the ACI paradigm the tones were not counter-balanced across positively and negatively valenced outcomes. The 2 kHz tone always predicted reward and 9 kHz tone always predicted punishment. This simplification has been made basing on our previous studies [14,21,23] and unpublished observations, which clearly show that animals are not biased toward any particular tone frequency. Second potential confound could be the use of tone cues in both behavioral tasks, which due to potential generalization could affect animals’ performance. This is however unlikely, because the tones used in the rSMT and ACI tests were clearly distinct (4 kHz and 15 kHz in the rSMT vs. 2 kHz and 9 kHz in the ACI). Similarly, although the ambiguouscues were not reinforced, the probability that animals could learn about this lack of contingency is negligible. The ambiguous-cues were relatively infrequent (10 out of 50 tones presented during single ACI session) and the ACI test sessions were performed in one week intervals what minimized mentioned above possibility. Moreover, analysis of the lever presses showed lack of differences in the proportion of omitted responses to the reference and ambiguous tones (2D). Another potential confound could be a possibility that animals exposed in the ACI paradigm to electric footshocks developed learned helplessness. This is however also unlikely as learned helplessness is a state in which an organism forced to bear aversive stimuli, becomes unable or unwilling to avoid subsequent encounters with those stimuli, even if they are ‘escapable’, because it has learned that it cannot control the situation. That was certainly not the case in our experimental design. During the training the animals learned to control the situation and to avoid unpleasant shock by pressing the ‘negative’ lever in answer to the tone predicting incoming punishment. As shown in Fig. 2C, the animals taking part in the experiment avoid successfully about 80% of shocks and remaining 20% are terminated after their onset. Moreover, because there were no significant differences between ‘optimists’ and ‘pessimists’ neither in the length of the negative tone training nor in the proportion of negative lever presses in response to the negative tone, it is unlikely that ‘pessimistic’ animals are more sensitive to footshocks or more likely to develop learned helplessness than their ‘optimistic’ counterparts. Finally, optimistic animals made more responses to ‘positive’ lever not only in response to the ambiguous-cues but also following the presentation of ‘positive’ reference tones, what might indicate a general increase in anticipation of reward. Although ideally, the differences in cognitive judgment bias should be indicated only by differences in responding to the ambiguous-cues, practice shows that often the bias is generalized and manifested also by altered responding to the positive and/or negative tones. Similar effects have been observed in our previous studies [14,15], and in the landmark study by Enkel et al. [10]. Last but not least, the comorbidity of gambling disorder with depression, at first sight, may seem problematic for interpretation of the present results. However, in our study we investigated interrelation between cognitive judgment bias and propensity to gamble in the healthy subjects. The ‘optimistic’ and ‘pessimistic’ judgment biases described in our study did not reflect pathological, depressive or maniacal states but were just behavioral traits of individual animals and as such could determine propensity of rats to certain aspects of gambling. Further studies should investigate
the co-morbidity of depression and gambling disorder in an animal models of depression e.g. based on chronic stress. 5. Conclusion In our recent studies, we have demonstrated a relationship between cognitive judgment bias and the vulnerability of animals to stress-induced anhedonia [14] and motivation [15]. The results of our present study revealed that, similar to humans, the cognitive judgment bias of rats could determine their propensity to gamble. Our research provides a unique platform for modeling interactions between behavioral traits and decision-making that indicate a risk for pathological behaviors in animals. Further studies should demonstrate whether a correlation within a (relatively small) healthy population could be extrapolated to a clinical sample. Conflict of interests The authors declare that there were no conflicts of interest in this study. Acknowledgments This work was supported by the Polish National Science Center (Research grant DEC-2014/13/B/NZ4/00214 to RR) and the statutory funds of the Institute of Pharmacology Polish Academy of Sciences. We thank Marcin Rzycki for graphical support. References [1] A. Blaszczynski, D. Silove, Cognitive and behavioral therapies for pathological gambling, J. Gambl. Stud. 11 (1995) 195–220. [2] T. Gilovich, Biased evaluation and persistence in gambling, J. Pers. Soc. Psychol. 44 (1983) 1110–1126. [3] R. Ladouceur, Gambling the hidden addiction, Can. J. Psychiatry 49 (2004) 501–503. [4] C.S. Carver, M.F. Scheier, Dispositional optimism, Trends Cogn. Sci. 18 (2014) 293–299. [5] B. Gibson, D.M. Sanbonmatsu, Optimism, pessimism, and gambling: the downside of optimism, Pers. Soc. Psychol. Bull. 30 (2004) 149–160. [6] M. Bateson, S. Desire, S.E. Gartside, G.A. Wright, Agitated honeybees exhibit pessimistic cognitive biases, Curr. Biol. 21 (2011) 1070–1073. [7] E.J. Bethell, A. Holmes, A. Maclarnon, S. Semple, Evidence that emotion mediates social attention in Rhesus macaques, PLoS One 7 (2007) e44387. [8] B.O. Brilot, L. Asher, M. Bateson, Stereotyping starlings are more pessimistic, Anim. Cogn. 13 (2010) 721–731. [9] R.E. Doyle, C. Lee, V. Deiss, A.D. Fisher, G.N. Hinch, A. Boissy, Measuring judgement bias and emotional reactivity in sheep following long-term exposure to unpredictable and aversive events, Physiol. Behav. 102 (2011) 503–510. [10] T. Enkel, D. Gholizadeh, O. von Bohlen Und Halbach, C. Sanchis-Segura, R. Hurlemann, R. Spanagel, et al., Ambiguous-cue interpretation is biased under stress- and depression-like states in rats, Neuropsychopharmacology 35 (2010) 1008–1015. [11] E.J. Harding, E.S. Paul, M. Mendl, Animal behaviour: cognitive bias and affective state, Nature 427 (2004) 312. [12] M. Mendl, J. Brooks, C. Basse, O. Burman, E. Paul, E. Blackwell, et al., Dogs showing separation-related behaviour exhibit a ‘pessimistic’ cognitive bias, Curr. Biol. 20 (2010) R839–R840. [13] R. Rygula, H. Pluta, P. Popik, Laughing rats are optimistic, PLoS One 7 (2012) e51959. [14] R. Rygula, J. Papciak, P. Popik, Trait pessimism predicts vulnerability to stress-induced anhedonia in rats, Neuropsychopharmacology 38 (2013) 2188–2196. [15] R. Rygula, J. Golebiowska, J. Kregiel, J. Kubik, P. Popik, Effects of optimism on motivation in rats, Front. Behav. Neurosci. 9 (2015) 32. [16] C.A. Winstanley, P.J. Cocker, R.D. Rogers, Dopamine modulates reward expectancy during performance of a slot machine task in rats: evidence for a ‘near-miss’ effect, Neuropsychopharmacology 36 (2011) 913–925. [17] C.A. Winstanley, Gambling rats: insight into impulsive and addictive behavior, Neuropsychopharmacology 36 (2011) 359. [18] P.J. Cocker, B. Le Foll, R.D. Rogers, C.A. Winstanley, A selective role for dopamine D(4) receptors in modulating reward expectancy in a rodent slot machine task, Biol. Psychiatry 75 (2014) 817–824.
D. Rafa et al. / Behavioural Brain Research 300 (2016) 97–105 [19] J. Papciak, P. Popik, E. Fuchs, R. Rygula, Chronic psychosocial stress makes rats more ‘pessimistic’ in the ambiguous-cue interpretation paradigm, Behav. Brain Res. 256 (2013) 305–310. [20] R. Rygula, J. Papciak, P. Popik, The effects of acute pharmacological stimulation of the 5-HT, NA and DA systems on the cognitive judgement bias of rats in the ambiguous-cue interpretation paradigm, Eur. Neuropsychopharmacol. 24 (2014) 1103–1111. [21] R. Rygula, E. Szczech, J. Papciak, A. Nikiforuk, P. Popik, The effects of cocaine and mazindol on the cognitive judgement bias of rats in the ambiguous-cue interpretation paradigm, Behav. Brain Res. 270 (2014) 206–212. [22] R. Rygula, J. Golebiowska, J. Kregiel, M. Holuj, P. Popik, Acute administration of lithium, but not valproate, modulates cognitive judgment bias in rats, Psychopharmacology (Berl.) 232 (2014) 2149–2156. [23] R. Rygula, E. Szczech, J. Kregiel, J. Golebiowska, J. Kubik, P. Popik, Cognitive judgment bias in the psychostimulant-induced model of mania in rats, Psychopharmacology (Berl.) 232 (2015) 651–660.
105
[24] D.C. Howell, Statistical Methods for Psychology, 4th ed., Wadsworth, Belmont, California, 1997. [25] D. Hecht, The neural basis of optimism and pessimism, Exp. Neurobiol. 22 (2013) 173–199. [26] A. Bari, D.E. Theobald, D. Caprioli, A.C. Mar, A. Aidoo-Micah, J.W. Dalley, et al., Serotonin modulates sensitivity to reward and negative feedback in a probabilistic reversal learning task in rats, Neuropsychopharmacology 35 (2010) 1290–1301. [27] T. Sharot, M. Guitart-Masip, C.W. Korn, R. Chowdhury, R.J. Dolan, How dopamine enhances an optimism bias in humans, Curr. Biol. 22 (2012) 1477–1481. [28] J. Kregiel, J. Golebiowska, P. Popik, R. Rygula, Dopamine induces an optimism bias in rats—pharmacological proof for the translational validity of the ambiguous-cue interpretation test, Behav. Brain Res. 297 (2015) 84–90.