Pigeons can helps us model gambling behavior; they too choose suboptimally, especially when they are hungry or isolated

An Animal Model of Human Gambling Behavior. Thomas R. Zentall. Current Research in Behavioral Sciences, February 26 2023, 100101. https://doi.org/10.1016/j.crbeha.2023.100101

Abstract: Human gambling almost always results in a loss. Thus, it is generally assumed that humans gamble for enjoyment and the enticement of winning. Although animals are purported to engage in optimal foraging behavior and should be sensitive to the probability of reinforcement, similar suboptimal behavior can be found in pigeons and other animals. They show a preference for an alternative that is associated with a signal for a low probability of a large reward (e.g., 20% probability of 10 pellets – a mean of 2 pellets) over an alternative that is associated with a signal for a high probability of a smaller reward (100% probability of 3 pellets). This effect may result from the strong conditioned reinforcement associated with a stimulus that is always followed by a reinforcer, but surprisingly, little conditioned inhibition associated with the signal for the absence of a reinforcer. A similar mechanism appears to be responsible for human gambling (gamblers tend to overvalue wins and undervalue losses). We have also found that for pigeons (and perhaps humans as well), the probability of the conditioned reinforcer is relatively unimportant, it is primarily the value of the reinforcer when it does occur (e.g., 10 pellets vs. 3 pellets) that is important. Interestingly, pigeons show several other parallels to human gambling behavior. For example, hungrier pigeons show a greater tendency to choose suboptimally. Also, pigeons that have had enrichment in the form of social experience in a larger cage show a reduced tendency to choose suboptimally. This animal model may provide a useful analog to human gambling behavior, one that is free from the influence of human culture, language, social reinforcement, and other experiential biases that may encourage human gambling.

Keywords: gamblingsuboptimal choiceimpulsivitypigeonshumans

Is Human gambling different from animal suboptimal choice?

The suboptimal choice procedures with pigeons differ somewhat from typical human gambling conditions. It could be argued, for example, that humans choose to bet with money that they already have, whereas pigeons choose between two alternatives, in either case the outcome of which is a gain. This means that for humans, a loss is a real loss, whereas for pigeons, it is a lost opportunity or an opportunity cost. In the case of human gambling, the choice is between a probable loss (the gamble) and no loss (refraining from the gamble), whereas for pigeons the suboptimal choice it is between less gain (suboptimal choice) and more gain (optimal choice). Thus, for the pigeon there is no actual cost. For this reason, however, one would think that for humans who gamble, an actual loss would be more aversive than a lost opportunity, and therefore, humans should be less likely to choose to gamble. In fact, as noted earlier, proportionally, humans are generally less likely to gamble (Molet et al., 2012) than pigeons are to choose suboptimally.

Second, in the pigeon task, there is typically a 10-s delay between the choice and the outcome of that choice (a reinforcer or its absence). However, for pigeons, it is not primarily the delay to the primary reinforcer that is critical, rather it is the immediacy of the appearance of the conditioned stimulus, and typically the conditioned stimulus appears immediately following the choice - if the conditioned stimulus does not appear immediately following choice, suboptimal choice is greatly reduced (McDevitt et al., 1997).

Third, the gambling task can be thought of as a go/no-go task (to gamble or not gamble), whereas the pigeon suboptimal choice task involves a simultaneous two-alternative choice. In fact, however, the human choice can be thought of as between the act of gambling and engaging in some other activity (e.g., going out to dinner or seeing a film). Thus, a clear distinction between the go/no-go task and the two-alternative choice task is not always easy to make.

One might suggest that another difference between the pigeon suboptimal choice task and human gambling is the importance of the post choice conditioned stimulus for the pigeon. If the signal for reinforcement is missing, pigeons generally choose optimally (Spetch et al., 1990; Spetch et al., 1994). Rats too chose risky options significantly more when win-associated audiovisual cues are added to the task (Barrus & Winstanley, 2016)

Although perhaps not as obvious as it is with pigeons, conditioned stimuli also play an important role in human gambling. For example, there are symbols that show up on the reels of a slot machine, numbers on the lottery ticket, and the ball falling into a winning number (or color) in roulette. If the only feedback gamblers received when they win or lose is receiving money or not, (imagine covering the symbols on a slot machine), it is unlikely that humans would be as inclined to gamble as they are. Evidence that conditioned stimuli play an important role for humans has been found by Cherkasova et al. (2018). Furthermore, Spetch et al. (2020) found that humans chose more risky slot machines when casino-related cues were associated with payouts.

Another difference between human gambling and pigeon suboptimal choice is that humans often describe gambling as an enjoyable activity (entertainment) rather than as a chance to make money. This difference is difficult to address directly because the emotion of a pleasurable activity is hard to measure (especially in pigeons). The question is, do humans gamble for the enjoyment of gambling rather than for the money? When it comes to games of chance, would humans be as likely to gamble if it were not possible to win money (or a prize)? The possibility of winning money (or the anticipation of the possibility of winning money) certainly contributes to what makes the game enjoyable. People may sometimes gamble for social rewards (being the best, earning the most points), but even with many social games (e.g., poker), humans often prefer to play for money rather than points alone.

I have suggested that the mechanism responsible for the pigeons’ suboptimal choice is the value of the outcome predicted by the conditioned stimulus, together with the difference between reward expected and reward obtained. An important component of this theory is the absence of inhibition-associated signals for the absence of reward. That is, the pigeons’ suboptimal choice is not inhibited by the frequent signals for the absence of reward. Consider the procedure used by Vasconcelos et al. (2015) in which starlings preferred the signaled alternative over the 50% unsignaled alternative. The signaled reinforcement was selected even when the signal for the absence of a reinforcer appeared 19 out of 20 times that the suboptimal alternative was selected, whereas reinforcement followed the unsignaled alternative half of the time. Thus, an important similarity between human gambling and pigeon suboptimal choice is the fact that, based on traditional reinforcement theory, losses contribute much less to conditioned inhibition than one might expect that they should, (Holst et al., 2010).

The cost of the relative absence of conditioned inhibition in the suboptimal choice procedure can be easily calculated. For pigeons, in the Case and Zentall (2018) experiment, choice of the optimal alternative would have resulted in twice as much food (50% vs. 100% reinforcement), in the Stagner and Zentall (2010) experiment, choice of the optimal alternative would have resulted in 2.5 times as much food (20 % vs. 50% reinforcement), and in the Vasconcelos et al, (2015) experiment, in the extreme condition, choice of the optimal alternative would have resulted in 10 times as much food (5% vs. 50% reinforcement). In these experiments, choice of the suboptimal alternative represents a considerably greater opportunity cost than the near-even chance of winning in many casino games (e.g., roulette or blackjack), although perhaps not as costly as the return on buying a lottery ticket.

Although the research with animals may appear to be different from human gambling, as noted earlier, not unlike pigeons, human gamblers are rarely aware of the probabilities of winning, but they are more likely to be aware of amount of money they would obtain if they won (billboards often announce the winning amount).

The research with pigeons suggests as well that there is an additional factor that may contribute to human gambling, positive contrast (or relative delay reduction), the positive feeling that gamblers may get when they expect a loss, but they experience a win. The results of the research on suboptimal choice with pigeons suggest that these tasks may provide a viable analog of unskilled gambling by humans. They may be useful in understanding the mechanisms responsible for human gambling behavior, as well as suggesting a possible treatment of problem gambling, the substitution of other pleasurable activities that do not have the negative social or economic implications that often accompany pathological gambling.