Global gain modulation generates time-dependent urgency during perceptual choice in humans

Abstract

Decision-makers must often balance the desire to accumulate information with the costs of protracted deliberation. Optimal, reward-maximizing decision-making can require dynamic adjustment of this speed/accuracy trade-off over the course of a single decision. However, it is unclear whether humans are capable of such time-dependent adjustments. Here, we identify several signatures of time-dependency in human perceptual decision-making and highlight their possible neural source. Behavioural and model-based analyses reveal that subjects respond to deadline-induced speed pressure by lowering their criterion on accumulated perceptual evidence as the deadline approaches. In the brain, this effect is reflected in evidence-independent urgency that pushes decision-related motor preparation signals closer to a fixed threshold. Moreover, we show that global modulation of neural gain, as indexed by task-related fluctuations in pupil diameter, is a plausible biophysical mechanism for the generation of this urgency. These findings establish context-sensitive time-dependency as a critical feature of human decision-making.

Neurodynamic Evidence Supports a Forced-Excursion Model of Decision-Making under Speed/Accuracy Instructions

“…The model-based behavioral literature suggests that a variation in the decision boundary (or, equivalently, a change in the baseline level) explains the SAT (Usher and McClelland, 2001; Smith and Ratcliff, 2004; Brown and Heathcote, 2008), but recent neural evidence has not supported this claim, suggesting more widespread changes (Heitz and Schall, 2012, 2013; Hanks et al, 2014; Murphy et al, 2016). To resolve this paradox, we hypothesized that the SAT may result from changes which are mathematically equivalent to a modulation of the decision boundary, but which are implemented physiologically through global changes in neural activity akin to turning up the gain in the brain.…”

“…In line with many previous studies (Usher and McClelland, 2001; Brown and Heathcote, 2008; Ratcliff and McKoon, 2008; Heitz, 2014), our free-excursion race model accounted for behavioral effects of the SAT, primarily by varying the amount of accumulated evidence required to make a decision. However, since recent studies exploring neural correlates of decision-making have challenged this implementation of the SAT (Heitz and Schall, 2012, 2013; Hanks et al, 2014; Murphy et al, 2016), we used a forced-excursion variant which models a global gain modulation by adjusting the parameters of the free-excursion race model so that the boundary was equal across SAT conditions, thus transferring the estimated difference between response bounds onto all other parameters affecting accumulation. In other words, a fixed boundary between SAT conditions was made mathematically equivalent to the free-excursion model by assuming different underlying mechanisms, with changes between SAT conditions explained not by boundary differences, but by differences between virtually all other parameters, modeling a global shift in decision-related brain activity.…”

“…Similarly, a recent human neuroimaging study proposed that urgency may arise from a global modulation of neural gain (Murphy et al, 2016). In fact, the concept of an evidence-independent urgency signal, which increases over time to inflate the accumulation process, has been a recurring theme in the recent SAT literature (Cisek et al, 2009; Milosavljevic et al, 2010; Thura et al, 2012).…”

An adaptive orienting theory of error processing

“…Moreover, transcutaneous vagus nerve stimulation, whichamong several other effects-may increase LC firing and NE release (Dietrich et al, 2008;Frangos, Ellrich, & Komisaruk, 2015), has also been found to increase PES (Sellaro et al, 2015). Furthermore, task-related modulations in pupil diameter are related to trial-by-trial adjustments of the speed-accuracy trade-off (Murphy, Boonstra, & Nieuwenhuis, 2016), reflecting the modulation of global neural gain associated with NE release (Aston-Jones & Cohen, 2005).…”

Neural Mechanisms for Executive Control of Speed-Accuracy Tradeoff

“…When viewed from the perspectives of formal models of SAT and previous studies of SAT using noninvasive measures of physiology (Forstmann et al, 2008Ivanoff et al, 2008;van Maanen et al, 2011;Murphy et al, 2016;Steinemann et al, 2018;van Veen et al, 2008;Weigard et al, 2018;Wenzlaff et al, 2011), coupled with prior extensive knowledge about the connectivity and functional properties of the brain circuits producing visually guided saccades (e.g., Liversedge et al, 2011), and previous studies of FEF (Heitz and Schall, 2012;Servant et al, 2019) and SC (Reppert et al, 2018), LIP (Hanks et al, 2014), skeletomotor cortex (Thura and Cisek, 2016) and basal ganglia (Thura and Cisek, 2017), the current findings offer several new insights into the neural mechanisms of SAT.…”