Responses to Modified Monty Hall Dilemmas in Capuchin Monkeys, Rhesus Macaques, and Humans

Abstract

The Monty Hall Dilemma (MHD) is a simple probability puzzle famous for its counterintuitive solution. Participants initially choose among three doors, one of which conceals a prize. A different door is opened and shown not to contain the prize. Participants are then asked whether they would like to stay with their original choice or switch to the other remaining door. Although switching doubles the chances of winning, people overwhelmingly choose to stay with their original choice. To assess how experience and the chance of winning affect decisions in the MHD, we used a comparative approach to test 264 college students, 24 capuchin monkeys, and 7 rhesus macaques on a nonverbal, computerized version of the game. Participants repeatedly experienced the outcome of their choices and we varied the chance of winning by changing the number of doors (three or eight). All species quickly and consistently switched doors, especially in the eight-door condition. After the computer task, we presented humans with the classic text version of the MHD to test whether they would generalize the successful switch strategy from the computer task. Instead, participants showed their characteristic tendency to stick with their pick, regardless of the number of doors. This disconnect between strategies in the classic version and a repeated nonverbal task with the same underlying probabilities may arise because they evoke different decision-making processes, such as explicit reasoning versus implicit learning.

Perceptions and Probabilities: Influence of Increased Options on Performance Generalization Across Two Variations of the Monty Hall Dilemma

“…Klein et al (2013) determined that the monkeys' (Macaca mulatta) performance indicated a substantial comparative similarity with human performance. The results of Watzek et al (2018) also showed considerable similarities between response patterns of capuchin (Cebus [Sapajus] apella) and rhesus macaques to human response patterns in both 3-door and 8-door conditions. Stagner and Zentall (2014) concluded that rats (Rattus norvegicus domestica, Sprague-Dawley strain), like pigeons, appeared to be influenced more by the overall probability of reinforcement, leading to similar performance between species.…”

“…Further studies examined whether this suboptimal tendency was exclusive to humans by testing nonhuman primates (Klein et al, 2013;Watzek et al, 2018) and rats (Stagner & Zentall, 2014). Klein et al (2013) determined that the monkeys' (Macaca mulatta) performance indicated a substantial comparative similarity with human performance.…”

“…Although research has demonstrated that human performance in the MHD can improve through repeated exposure, typical performance is still well below optimal. Several studies increased the number of options available for selection to provide additional performance improvements (Burns & Wieth, 2004;Franco-Watkins et al, 2003;Saenen et al, 2015;Stibel et al, 2009;Watzek et al, 2018;reviewed by Saenen et al, 2018). Increasing the number of available options raises the likelihood that switching will result in reinforcement when all but two options have been eliminated.…”

Humans’ (Homo sapiens), capuchin monkeys’ (Sapajus [Cebus] apella), and rhesus macaques’ (Macaca mulatta) size judgments shift when stimuli change in frequency.

“…Further, tasks are modified to be presented to the animals (at bare minimum, they have to be adjusted to provide training without verbal instruction) and at times, when humans are “back tested” on the modified task, they have shown different responses than in the original task; if the humans had not been tested on the modified task, we would have reached the erroneous conclusion that the nonhumans behaved differently than the humans. As an example of this, when given the same modified computer task to simulate the Monty Hall Dilemma, humans, similarly to capuchins and rhesus monkeys, quickly and consistently chose to switch doors, which is contrary to humans’ results when they were given the task with verbal instructions (Watzek et al, 2018). Thus, comparative studies require that all species perform the modified task by using the same (or as similar as possible) methodologies, including (or especially) humans (Smith et al, 2018).…”

“…In this study, our goal was to determine whether capuchin and rhesus monkeys would expand their concepts of items that became less frequent over time using a methodology drawn from the human literature, and to directly compare the monkeys’ choices to humans’ by running human participants on the same task. Both monkey species are susceptible to some of the same decision-making biases as humans in some contexts that favor relative rather than absolute judgments (Chen et al, 2006; Egan Brad et al, 2016; Lakshminarayanan et al, 2008, 2011; Santos & Rosati, 2015; Watzek & Brosnan, 2020a, 2020b, 2020c), although this is not always the case (Cohen & Santos, 2017; Watzek & Brosnan, 2018; Watzek et al, 2018, 2019; Xu et al, 2011). For example, there is a tendency for both humans and monkeys to overvalue objects that they own over objects that they do not yet own, also known as the endowment effect (Lakshminarayanan et al, 2008).…”