Opposite effects of choice history and evidence history resolve a paradox of sequential choice bias

Abstract

Perceptual decisions are biased toward previous decisions. Earlier research suggests that this choice repetition bias is increased after previous decisions of high confidence, as inferred from response time measures (Urai, Braun, & Donner, 2017), but also when previous decisions were based on weak sensory evidence (Akaishi, Umeda, Nagase, & Sakai, 2014). As weak sensory evidence is typically associated with low confidence, these previous findings appear conflicting. To resolve this conflict, we set out to investigate the effect of decision confidence on choice repetition more directly by measuring explicit confidence ratings in a motion coherence discrimination task. Moreover, we explored how choice and evidence history jointly affect subsequent perceptual choices. We found that participants were more likely to repeat previous choices of high subjective confidence, as well as previous fast choices, confirming the boost of choice repetition with decision confidence. Furthermore, we discovered that current choices were biased away from the previous evidence direction and that this effect grew with previous evidence strength. These findings point toward simultaneous biases of choice repetition, modulated by decision confidence, and evidence adaptation, modulated by the strength of evidence, which bias current perceptual decisions in opposite directions.

Perceptual history propagates down to early levels of sensory analysis

“…Although the mechanisms mediating serial dependence remain far from clear, our findings with orientation judgments point to a relatively early site of action of the priors, a site encoding raw orientation signals unaffected by spatial context. Obviously, this does not exclude the role of other factors, like attention 10,29 and confidence, [30][31][32] or processes such as short-term memory. 16,[33][34][35][36] How these interact with the early mechanisms revealed in this study remains open, subject to further research.…”

Increased influence of prior choices on perceptual decisions in autism

“…A positive influence of prior choices has also been documented in many studies ( Kaneko and Sakai, 2015 ; Talluri et al, 2018 ; Feigin et al, 2021b ). Furthermore, converging evidence from many recent studies suggests that the influence of prior stimuli and prior choices are often in opposite directions – that is, a negative (adaptive) influence of prior stimuli vs. a positive (consistency bias) influence of prior choices, and that these superimpose ( Fritsche et al, 2017 ; Bosch et al, 2020 ; Feigin et al, 2021b ; Sadil et al, 2021 ). However, opposing effects of prior sensory vs. prior choices on perceptual decisions has not yet been studied in ASD.…”

Same Stimulus, Same Temporal Context, Different Percept? Individual Differences in Hysteresis and Adaptation When Perceiving Multistable Dot Lattices

“…Although earlier research has found evidence for individual differences in several tasks assessing hysteresis, adaptation, or their ratio (e.g., Abrahamyan, Silva, Dakin, Carandini, & Gardner, 2016;Mattar, Carter, Zebrowitz, Thompson-Schill, & Aguirre, 2018;McGovern, Walsh, Bell, & Newell, 2017;, only a few studies have attempted to quantify both effects concurrently at the level of the individual participant, by distinguishing the effects of previous stimulus support and previous percept or response (e.g., Bosch, Fritsche, Ehinger, & de Lange, 2020;Urai, Braun, & Donner, 2017;Zhang & Alais, 2019). Moreover, in none of these studies individual differences in effects of previous stimulus and perceptual choice were the focus of study.…”

“…Although almost all participants showed an attractive choice history bias and all participants showed a repulsive evidence history bias, the size of the choice history bias varied considerably across participants (cf. Supplementary Figure 2 in Bosch, Fritsche, Ehinger, & de Lange, 2020).…”

“…Additional arguments for assuming separate mechanisms are differences in the extent to which hysteresis and adaptation are dependent on attention, are modulated by subjective confidence, are modulated by working memory delay, or exhibit clear spatial specificity (for an overview, see Fritsche, Spaak, & de Lange, 2020). Many have also distinguished the effects based on their source being stimulus-related, percept-related, choice-related or motor-related (e.g., Bosch, Fritsche, Ehinger, & de Lange, 2020;Carter, Snyder, Fung, & Rubin, 2014;Cicchini, Mikellidou, & Burr, 2017;Pascucci et al, 2019;Sadil, Cowell, & Huber, 2021;Zhang & Alais, 2019).…”