A pupillary index of susceptibility to decision biases

Abstract

Under what conditions do humans systematically deviate from rational decision making? Here we show that pupillary indices of low neural gain are associated with strong and consistent biases across six different extensively-studied decision making tasks, whereas indices of high gain are associated with weak or absent biases. Lower susceptibility to biases, however, comes at the cost of indecisiveness, or alternatively, prolonged deliberation time. We explain the association between low gain and strong biases as reflecting a broader information integration process that gives greater weight to weak biasing influences. The findings underscore the role of pupil-linked brain states in the generation of decision making biases.

Correlated P300b and phasic pupil‐dilation responses to motivationally significant stimuli

“…This may seem incompatible with the possibility of discarding an action-related component from the autonomic (pupillary) response, leaving a component that still meaningfully relates to the different stimulus types. However, this is in line with accumulating evidence that pupil dilations are not only linked with actions but also reliably index aspects of stimulus processing, particularly linked with their match to expectations and prior beliefs (de Gee et al, 2017;Eldar et al, 2021;Murphy et al, 2021). In turn, these observations are coherent with the proposal that phasic pupil dilations are often (Joshi et al, 2016;Reimer et al, 2016) though not always (Megemont et al, 2022) linked with activity in the locus coeruleus and the consequent diffuse release of NE in many cortical regions.…”

Teleological reasoning bias is predicted by pupil dynamics: Evidence for the extensive integration account of bias in reasoning

“…However, a single-process framework, the Extensive integration account, makes the opposite prediction, namely that bias in reasoning is exacerbated by more extensive processing. Recently, Eldar et al (2021) highlighted that dual-process theories and the Extensive integration account make opposing predictions regarding pupil dilation and found support for the Extensive integration account in three framing tasks.…”

“…Recently, Eldar et al. (2021) highlighted that dual‐process theories and the Extensive integration account make opposing predictions regarding pupil dilation and found support for the Extensive integration account in three framing tasks.…”

“…Thus, a small bias will have larger effects if each piece of evidence has lower weighting and the decision requires a longer evidence accumulation process. Thus, more extensive integration is associated with more bias (Eldar et al., 2021; Usher & McClelland, 2004). Importantly, it is proposed that evidence integration is influenced by the Locus Coeruleus–Norepinephrine system, as norepinephrine modulates neural gain (Aston‐Jones & Cohen, 2005; Eldar et al., 2013; Eldar, Cohen, et al., 2016; Eldar, Niv, et al., 2016; Jepma & Nieuwenhuis, 2011; Joshi et al., 2016).…”

“…Importantly, it is proposed that evidence integration is influenced by the Locus Coeruleus–Norepinephrine system, as norepinephrine modulates neural gain (Aston‐Jones & Cohen, 2005; Eldar et al., 2013; Eldar, Cohen, et al., 2016; Eldar, Niv, et al., 2016; Jepma & Nieuwenhuis, 2011; Joshi et al., 2016). Low neural gain leads to lower weighting of each piece of evidence, and thus more extensive integration is required to reach a decision (Eldar, Cohen, et al., 2016; Eldar et al., 2013, 2021). Conversely, high neural gain leads to increased weighting of each piece of evidence.…”