Research has shown that pigeons given a simultaneous visually based discrimination reversal, in which a single reversal occurs at the midpoint of each session, consistently show anticipation prior to the reversal as well as perseveration after the reversal, suggesting that they use a less effective cue (time or trial number into the session) than what would be optimal to maximize reinforcement (local feedback from the most recent trials). In the present research, pigeons (Columba livia) and rats (Rattus norvegicus) were tested with a simultaneous spatial discrimination midsession reversal. Pigeons showed remarkably similar errors in anticipation and perseveration as with visual stimuli, thereby continuing to show the suboptimal use of time as a cue, whereas rats showed no anticipatory errors and very few perseverative errors, suggesting that they used local feedback as a cue, thus more nearly optimizing reinforcement. To further test the rats' flexibility, they were then tested with a variable point of reversal and then with multiple points of reversal within a session. Results showed that the rats effectively maximized reinforcement by developing an approximation to a win-stay/lose-shift rule. The greater efficiency shown by rats with this task suggests that they are better able to use the feedback from their preceding choice as the basis of their future choice. The difference in cue preference further suggests a qualitative difference in acquisition of the midsession reversal task between pigeons and rats.

Consistent with human gambling behavior but contrary to optimal foraging theory, pigeons show a strong preference for an alternative with low probability and high payoff (a gambling-like alternative) over an alternative with a greater net payoff (Zentall & Stagner, Proceedings of the Royal Society B, 278, 1203Society B, 278, -1208Society B, 278, , 2011. In the present research, we asked whether humans would show suboptimal choice on a task involving choices with probabilities similar to those for pigeons. In Experiment 1, when we selected participants on the basis of their self-reported gambling activities, we found a significantly greater choice of the alternative involving low probability and high payoff (gambling-like alternative) than for a group that reported an absence of gambling activity. In Experiment 2, we found that when the inhibiting abilities of typical humans were impaired by a self-regulatory depletion manipulation, they were more likely to choose the gambling-like alternative. Taken together, the results suggest that this task is suitable for the comparative study of suboptimal decision-making behavior and the mechanisms that underlie it.Keywords decision making . suboptimal choice . gambling . humans . pigeons A subset of suboptimal decision making can be characterized as making a decision to choose a low-probability but high-payoff alternative (e.g., playing slot machines or buying lottery tickets) over a high-probability, low-payoff alternative (not gambling), such that the net expected return is less than what one has wagered. In this kind of gambling, over the long term, one is very likely to lose more than one wins, and having sufficient experience with the game to learn about the odds of winning does not appear to reduce its frequency. It appears that gamblers tend to over-attend to wins and discount losses (the availability heuristic; Tversky & Kahneman, 1974).Recently, Zentall and colleagues studied suboptimal decision-making processes in pigeons (Gipson, Alessandri, Miller, & Zentall, 2009;Stagner & Zentall, 2010;Zentall & Stagner, 2011). They asked whether pigeons would show choice behavior analogous to the suboptimal behavior shown by humans when humans purchase a lottery ticket or engage in casino gambling. To answer this question, Zentall and Stagner designed a task that may be considered analogous to human gambling (see Fig. 1a). In this task, pigeons are given a choice between one alternative reliably associated with three reinforcers (analogous to money in one's pocket) and a second alternative (analogous to gambling) associated with a low-probability, high-payoff stimulus on some trials (10 pellets on 20% of the trials) or a highprobability, zero-payoff stimulus on others. In one sense, the task is analogous to suboptimal human gambling, because the average payoff for the first alternative (with an overall mean reinforcement per trial of 3 pellets) is greater than that of the second alternative (with an overall mean reinforcement per trial of 2 pellets). Results show...

Pigeons (Columba livia) produce many anticipatory and perseverative errors on discrimination tasks with a reversal of reward contingencies partway through the session. Prior comparative research has suggested that rats (Rattus norvegicus) do not show the same number of errors and produce results that more closely resemble those of humans. We examined pigeons' performance on a visual-spatial discrimination with the reversal point randomized within the session and found that they showed remarkably few errors. When these subjects were split into groups with the contingencies for reward unconfounded, the birds in the spatial-contingency group maintained their performance, and those in the visual-contingency group made many more anticipatory and perseverative errors. We also examined the performance of naïve pigeons on a spatial midsession reversal task and found a pattern of results similar to those shown by pigeons that had previously been trained on a visual-spatial reversal procedure. Finally, we studied rats on a T-maze using a spatial-discrimination midsession reversal task and found that the rats showed a large number of anticipatory and perseverative errors. Near-perfect performance on the midsession reversal task appears to be subject to the ability of the animal to orient spatially during the intertrial interval, rather than being due to broad species differences.

In 2 experiments, rats were trained to press a centrally located lever that delivered immediate food reinforcement and turned on a light signal that indicated the location of a further food reward. After rats learned to press the lever and use the light cue to find food, immediate reinforcement for lever pressing was discontinued. In Experiment 1, rats continued to press the lever for information about the location of reward in a T-maze, but control groups yoked to the experimental group for amount of reward, and conditioned reinforcement showed complete extinction of lever pressing. Rats tested on an 8-arm radial maze in Experiment 2 also continued to press a lever that did not yield immediate reinforcement but provided a light cue indicating which randomly chosen arm of the maze contained food; lever pressing declined significantly, however, when the same arm contained food on every trial. Comparisons of testing conditions between and within experiments suggested that probability of lever pressing increased as the amount of information gained increased. The comparative implications of these findings for metacognition are discussed.

No abstract