Quantifying the immediate computational effects of preceding outcomes on subsequent risky choices

Abstract

Forty years ago, prospect theory introduced the notion that risky options are evaluated relative to their recent context, causing a significant shift in the study of risky monetary decision-making in psychology, economics, and neuroscience. Despite the central role of past experiences, it remains unclear whether, how, and how much past experiences quantitatively influence risky monetary choices moment-tomoment in a nominally learning-free setting. We analyzed a large dataset of risky monetary choices with trial-by-trial feedback to quantify how past experiences, or recent events, influence risky choice behavior and the underlying processes. We found larger recent outcomes both negatively influence subsequent risk-taking and positively influence the weight put on potential losses. Using a hierarchical Bayesian framework to fit a modified version of prospect theory, we demonstrated that the same risks will be evaluated differently given different past experiences. The computations underlying risky decision-making are fundamentally dynamic, even if the environment is not. Prospect theory is conceivably the most successful recent theory in risky decision-making 1. The theory is founded on the idea that people evaluate options relative to "the past and present context of experience" (p. 277) 1. While this central insight led to two of the most successful components of prospect theory, the reflection effect in risk aversion (diminishing marginal utility of values expressed as concavity for gains and convexity for losses) and loss aversion (overweighting losses compared to gains of the same magnitude), it is unclear how recent events quantitatively influence risky monetary decision-making, moment-to-moment. We sought to quantify how recent events (e.g. previous outcomes) influence choices by analyzing a large dataset of risky choices in the presence of feedback compiled from four studies 2-5. Laboratory and field studies have suggested that recent events shape the subjective value of risky choice options 6-9 and actions 10. The directionality and mechanism supporting these effects are unclear as some studies of risky decision-making in the presence of feedback have found that previous gains, relative to losses, lead to less risk-taking 6,10 or a mixture of less and more risk-taking 7-9,11 , discrepancies that may be related to how and when outcomes are realized, or paid 12. Theories that have sought to specify the directionality of the effect of feedback on risky choice have not described the size of the effect, its characteristics (e.g. how far it extends in time) or the precise decision processes that would be altered by feedback 13. Interpreting these previous findings is complicated by the fact that almost all of these studies included design features intended to elicit temporal context effects (e.g. explicitly-signaled contexts, displays with cumulative elements) and thus may have strengthened, or even created, dynamic context effects. While such an approach identifies an upper bound on the possible presence and ma...

Choosing for others changes dissociable computations underlying risky decision-making

“…We sought to characterize the separable cognitive mechanisms by which social context affected decision computations in two models of risky decision-making based on prospect theory (Kahneman & Tversky, 1979;Tversky & Kahneman, 1992): Model 1 and Model 2. (Brooks & Sokol-Hessner, 2020;Sokol-Hessner et al, 2016), and similar to extant computational work (Scheibehenne & Pachur, 2015; see also Methods and Supplementary Materials), these additive change terms are not directly in 'parameter' (i.e., ⍴, λ, and µ) space. Their effect size can nevertheless be understood in terms of how they change the mean 'self'…”

“…Baseline valuation processes were modeled in a manner similar to that described in prior work (Brooks & Sokol-Hessner, 2020;Scheibehenne & Pachur, 2015;Sokol-Hessner et al, 2016). We first computed the utility associated with the potential gain ( # ) and loss ( ' ) outcomes of a gamble on a given trial using Equations 2 and 3 respectively:…”

Between Joy and Sympathy: Smiling and Sad Recipient Faces Increase Prosocial Behavior in the Dictator Game

“…In addition, as we had no explicit hypothesis and only 20 trials with five different emotional states, we did not analyze the influence of preceding decisions and identities that occurred more than one round earlier, which may be a valuable adaption in future studies. As suggested in recent research [ 47 ], economic decisions are not necessarily independent of each other, a finding that has so far been of secondary interest.…”

“…Moreover, we expected a main effect of trait EC on prosocial behavior [ 41 ] as well as an interaction between trait EC and sad faces, as these faces might elicit sympathy in the dictator [ 25 ]. Building on previous literature on prosocial habits and trial-by-trial analyses [ 47 , 48 ], we examined whether the face of the preceding recipient and the participants’ preceding decision might influence decision-making in the current trial.…”

Early Caregiving Adversity Differentially Shapes Behavioral Sensitivity to Reward and Risk during Decision-Making

“…Our analysis consisted of two overarching approaches. First, we modeled task responses at the trial level (i.e., likelihood of making a risky decision) to understand the effects that taskrelated and between-person variables had on the likelihood of risk-taking (Brooks & Sokol-Hessner, 2020). Next, we used formal computational models to better understand the mechanistic underpinnings (loss and risk aversion, choice consistency, target threshold) of task responses (Navarro, Tran, & Baz, 2018).…”