Flexible categorization in perceptual decision making

Abstract

Perceptual decisions rely on accumulating sensory evidence. This computation has been studied using either drift diffusion models or neurobiological network models exhibiting winner-take-all attractor dynamics. Although both models can account for a large amount of data, it remains unclear whether their dynamics are qualitatively equivalent. Here we show that in the attractor model, but not in the drift diffusion model, an increase in the stimulus fluctuations or the stimulus duration promotes transitions between decision states. The increase in the number of transitions leads to a crossover between weighting mostly early evidence (primacy) to weighting late evidence (recency), a prediction we validate with psychophysical data. Between these two limiting cases, we found a novel flexible categorization regime, in which fluctuations can reverse initially-incorrect categorizations. This reversal asymmetry results in a non-monotonic psychometric curve, a distinctive feature of the attractor model. Our findings point to correcting decision reversals as an important feature of perceptual decision making.

A confirmation bias in perceptual decision-making due to hierarchical approximate inference

“…Each trial, the Extended ITB model followed the noisy integration dynamics in Eq (22), where LPO 0 ¼ log p C 1À p C and LLO f was computed exactly, as described above. After integration, the decision then incorporated a symmetric lapse rate and temperature: pðChoice ¼ þ1jLPO F ; l; TÞ ¼ l þ ð1 À 2lÞs ðLPO F =TÞ ;…”

“…The analysis by Glaze et al (2015) shows that a recency bias is optimal in a volatile environment, but such mechanisms cannot explain primacy effects [14]. Deneve (2012)'s normative analysis predicts that primacy and recency should depend on trial-by-trial changes in difficulty [21], while Prat-Ortega et al (2021) find that primacy and recency can change as a function of the variability of the input to a attractor-based decision-circuit [22]. However, neither account alone, or in combination, can explain the differences found across experiments.…”

Bump attractor dynamics underlying stimulus integration in perceptual estimation tasks

“…Mathematically, the bump attractor model can be viewed as a generalization of the discrete attractor model of perceptual decision making 8,9,11,54 . However, the dynamics of evidence integration in the two models are fundamentally different.…”

“…Discrete attractor models have a double-well potential that usually leads to a primacy temporal weighting because once the system settles into one of the two attractors it remains there until the end of the trial. Recently we have described how uniform and recency weighting can be obtained when fluctuations in the stimulus together with the internal noise are strong enough to overcome the categorization dynamics and cause transitions between the attractor states 54 . In contrast to this, in the bump attractor model a continuous integration process -without abrupt transitions -occurs in the neutrally stable ring manifold and the stimulus always impacts the bump phase, yielding recency without the need for strong fluctuations (Fig.…”

“…For the model simulations, we obtain the PK as the stimulus weights of a circular regression model. To quantify the shape of a PK, we define the PK slope 54 . It is the slope of a linear regression of the PK, normalized between −1 (decaying PK, primacy) and +1 (increasing PK, recency).…”