Adaptable history biases in human perceptual decisions

Abstract

When making choices under conditions of perceptual uncertainty, past experience can play a vital role. However, it can also lead to biases that worsen decisions. Consistent with previous observations, we found that human choices are influenced by the success or failure of past choices even in a standard two-alternative detection task, where choice history is irrelevant. The typical bias was one that made the subject switch choices after a failure. These choice history biases led to poorer performance and were similar for observers in different countries. They were well captured by a simple logistic regression model that had been previously applied to describe psychophysical performance in mice. Such irrational biases seem at odds with the principles of reinforcement learning, which would predict exquisite adaptability to choice history. We therefore asked whether subjects could adapt their irrational biases following changes in trial order statistics. Adaptability was strong in the direction that confirmed a subject’s default biases, but weaker in the opposite direction, so that existing biases could not be eradicated. We conclude that humans can adapt choice history biases, but cannot easily overcome existing biases even if irrational in the current context: adaptation is more sensitive to confirmatory than contradictory statistics.

Pupil-linked arousal is driven by decision uncertainty and alters serial choice bias

“…Although such serial choice biases were first identified in psychophysical tasks about a century ago32, their determinants have remained poorly understood. Previous treatments of serial choice biases have conceptualized experimental history as sequences of binary external events (past stimulus identities, choices, or feedback)3339. We here established that these serial biases were also modulated by the decision-maker's pupil-linked arousal state on the previous trial, which, in turn, reflected the uncertainty about the observer's choice.…”

Response outcomes gate the impact of expectations on perceptual decisions

“…4a). This win-stay-lose-switch strategy has been extensively characterized both in humans (Abrahamyan et al, 2016;Braun et al, 2018;Fründ et al, 2014) and rodents (Akrami et al, 2018;Hwang et al, 2017) . Second, we identified the sequential transition bias, a form of rule bias that had been previously shown to impact human reaction times (Cho et al, 2002;Kirby, 1976;Soetens et al, 1985) , choices (Maloney et al, 2005) and neural responses (Jones et al, 2013;Sommer et al, 1999) .…”

“…To quantify the impact onto current decisions of the lateral and the transitions biases, and to investigate their dependence on error responses, we used a generalized linear model (GLM) (Abrahamyan et al, 2016;Braun et al, 2018;Busse et al, 2011;Fründ et al, 2014;Nogueira et al, 2017;Urai et al, 2017) . The GLM separately measured the impact onto the current decision of each response r ( r = R, L ) and transition T ( T = Rep, Alt ) in the last ten trials ( Supplementary Fig.…”

“…In dynamical environments, in which the statistics of the sensory information varies with time, subjects must be constantly updating their internal model by accumulating past stimuli, actions and outcomes (Yu and Cohen, 2008) . The updating of the prior based on the actions occurring in each trial typically introduces sequential effects, which are systematic history-dependent choice biases reflecting the impact of the trial-to-trial variations in expectation (Abrahamyan et al, 2016;Akaishi et al, 2014;Akrami et al, 2018;Ashourian and Loewenstein, 2011;Braun et al, 2018;Busse et al, 2011;Cho et al, 2002;Fischer and Whitney, 2014;Fründ et al, 2014;Hwang et al, 2017;Meyniel et al, 2016;Nogueira et al, 2017) . However, there are circumstances where subjects seem able to quickly and flexibly modulate the impact of prior expectations in driving their choices.…”

Regulation of evidence accumulation by pupil-linked arousal processes

“…Thus, when making multiple perceptual decisions, the current decision is influenced by the choice we just made, for example by repeating an action when it was rewarded and choosing something else when it was not, a reinforcement learning e ect (Abrahamyan, Silva, Dakin, Carandini, & Gardner, 2016;Daw et al, 2006;Drugowitsch et al, 2016;Sutton & Barto, 1998), or simply repeating a choice regardless of the outcome associated with it, a choice kernel e ect (Abrahamyan et al, 2016;Drugowitsch et al, 2016). Such sequential dependence can be advantageous when there are temporal correlations between trials, as is the case in many reinforcement learning tasks (Daw et al, 2006;Sutton & Barto, 1998), but is suboptimal in most perceptual decision making tasks when each trial is independent of the past (Abrahamyan et al, 2016;Akrami, Kopec, Diamond, & Brody, 2018;Barraclough, Conroy, & Lee, 2004;Drugowitsch et al, 2016).…”