Utilizing Bayesian ideal-observer models, we have shown rational inference and decision-making naturally give rise to apparently sub-optimal behavior in a variety of classical psychology paradigms, including the widely observed tendency of humans to show apparently superstitious sequential effects in 2-alternative forced choice tasks (Yu & Cohen, 2009; Ma & Yu, 2015; Ryali & Yu, 2018), to allocate choice preferences using matching rather than maximizing (optimal) in active sensing (Yu & Huang, 2014), to exhibit grouping effects in visual crowding (Zhang, Song, & Yu, 2015), and to alter relative preferences for consumer products in the presence/absence of other options (Shenoy & Yu, 2013).

Kahneman and Tversky's Prospect Theory

The last of these phenomena contributed to Kahneman and Tversky’s Prospect Theory, which explains human decision-making behavior largely in terms of heuristics and biases; our work showed that there is no need to invoke an ad hoc euristic explanation, since such contextual effects are natural consequences of using the available options as a means of gaining information about the market environment.

We have also found that humans are systematically devalue unchosen options and over-estimate environmental volatility, in the multi-armed bandit task, but the two suboptimalities together, along with a simplistic softmax decision policy, compensate for one another and result in near optimal performance (Guo & Yu, 2018), an insight that might also prove useful for guiding AI systems to efficiently negotiate exploration-exploitation tradeoffs.

• Yu, A J & Cohen, J D (2009). Sequential effects: Superstition or rational behavior? Advances in Neural Information Processing Systems 21: 1873-1880.
• Shenoy, P & Yu, A J (2013). A rational account of contextual effects in preference choice: What makes for a bargain? Proceedings of the Thirty-Fifth Annual Conference of the Cognitive Science Society.
• Yu, A J & Huang, H (2014). Maximizing masquerading as matching: Statistical learning and decision-making in choice behavior. Decision, 1 (4): 275-287.
• Zhang, S, Song, M, & Yu, A J (2015). Bayesian hierarchical model of local-global processing: Visual crowding as a case-study. Proceedings of the Cognitive Science Society Conference.
• Ma, N & Yu, A J (2015). Statistical learning and adaptive decision-making underlie human response time variability in inhibitory control. Frontiers in Psychology, 6 (1046).
• Ryali, C, Reddy, G, Yu, A J (2018). Demystifying excessively volatile human learning: A Bayesian persistent prior and a neural approximation. Advances in Neural Information Processing Systems. Cambridge, MA: MIT Press.
• Guo, D, Yu, A J (2018). Why so gloomy? A Bayesian explanation of human pessimism bias in the multi-armed bandit task. Advances in Neural Information Processing Systems. Cambridge, MA: MIT Press.