Comparing utility functions between risky and riskless choice in rhesus monkeys

Abstract

Decisions can be risky or riskless, depending on the outcomes of the choice. Expected Utility Theory describes risky choices as a utility maximization process: we choose the option with the highest subjective value (utility), which we compute considering both the option’s value and its associated risk. According to the random utility maximization framework, riskless choices could also be based on a utility measure. Neuronal mechanisms of utility-based choice may thus be common to both risky and riskless choices. This assumption would require the existence of a utility function that accounts for both risky and riskless decisions. Here, we investigated whether the choice behavior of macaque monkeys in riskless and risky decisions could be described by a common underlying utility function. We found that the utility functions elicited in the two choice scenarios were different from each other, even after taking into account the contribution of subjective probability weighting. Our results suggest that distinct utility representations exist for riskless and risky choices, which could reflect distinct neuronal representations of the utility quantities, or distinct brain mechanisms for risky and riskless choices. The different utility functions should be taken into account in neuronal investigations of utility-based choice.

Reward value revealed by auction in rhesus monkeys

“…To express subjective value properly in utils would have required estimation of a utility function for water separately for each animal. While our monkeys show different degrees of nonlinearity in riskless and risky utility functions for various liquids (Stauffer et al, 2014;Genest et al, 2016;Bujold et al, 2021), the current absence of an empirically estimated utility function for water would render the utility of the measured bids basically uninformative. Nevertheless, we aimed to obtain supplementary confirmatory information on the bid-juice relationship by using other models, including a linear function, two power functions and a 2-parameter Prelec function (Prelec, 1998).…”

Systematic comparison of risky choices in humans and monkeys

“…Risky decision making has been investigated for more than 200 years (Stigler, 1950) and remains a popular research topic (Blavatskyy et al, 2022;Bujold et al, 2022;Frey et al, 2017;Ruggeri et al, 2020;Yang et al, 2022). While economists showed interest in building mathematical models to explain risky decisions and relating them to society (Moscati, 2016;Ruggeri et al, 2020;Schneider & Day, 2018), neuroscientists investigated the neural mechanism of decision making.…”

Risky choice: probability weighting explains Independence Axiom violations in monkeys

“…Their reliable and stable performance minimizes errors and rules out insufficient learning, as noted for rodent tests of the IA axiom (Battalio et al, 1985;Kagel et al, 1990). Monkeys' choices satisfy first-, second-and third-order stochastic dominance, allow comparisons between risky and riskless utility functions, reveal nonlinear probability weighting, comply with the EUT continuity axiom, and can respect the Independence of Irrelevant Alternatives (IIA) of two-component bundles (Stauffer et al, 2014;Genest et al, 2016;Pastor-Bernier et al, 2017;Ferrari-Toniolo et al, 2019;Bujold et al, 2021). However, without testing the IA, these choice data do not yet allow us to identify specific forms of subjective value computation.…”

“…Their reliable and stable performance minimizes errors and rules out insufficient learning, as noted for rodent tests of the IA axiom (Camerer, 1989; Kagel, Macdonald, Battalio, Kagel, & Mac, 1990). Monkeys’ choices satisfy first-, second- and third-order stochastic dominance, allow comparisons between risky and riskless utility functions, reveal nonlinear probability weighting, comply with the EUT continuity axiom, and can respect the Independence of Irrelevant Alternatives of two-component bundles (Bujold, Seak, Schultz, & Ferrari-Toniolo, 2021; Ferrari-Toniolo, Bujold, Grabenhorst, Báez-Mendoza, & Schultz, 2021; Ferrari-Toniolo, Bujold, & Schultz, 2019; Genest, Stauffer, & Schultz, 2016; Pastor-Bernier, Plott, & Schultz, 2017; Pelé, Broihanne, Thierry, Call, & Dufour, 2014; Stauffer, Lak, Bossaerts, & Schultz, 2015; Stauffer, Lak, & Schultz, 2014). However, without testing the IA, these choice data do not yet allow us to identify specific forms of subjective value computation.…”