Gazing at Social Interactions Between Foraging and Decision Theory

Abstract

Finding the underlying principles of social attention in humans seems to be essential for the design of the interaction between natural and artificial agents. Here, we focus on the computational modeling of gaze dynamics as exhibited by humans when perceiving socially relevant multimodal information. The audio-visual landscape of social interactions is distilled into a number of multimodal patches that convey different social value, and we work under the general frame of foraging as a tradeoff between local patch exploitation and landscape exploration. We show that the spatio-temporal dynamics of gaze shifts can be parsimoniously described by Langevin-type stochastic differential equations triggering a decision equation over time. In particular, value-based patch choice and handling is reduced to a simple multi-alternative perceptual decision making that relies on a race-to-threshold between independent continuous-time perceptual evidence integrators, each integrator being associated with a patch.

Inferring Causal Factors of Core Affect Dynamics on Social Participation through the Lens of the Observer

“…Interestingly enough, the first stage of the emulator could be exploited whenever an analysis in terms of stochastic trajectories is requested. One straightforward example is the analysis of gaze trajectories in social/affective behaviour [ 94 , 95 ].…”

Pain and Fear in the Eyes: Gaze Dynamics Predicts Social Anxiety from Fear Generalisation

“…As to the analyses, a number of works have recently considered eye movements modelling from the foraging perspective (e.g., [23,8,7,22,20,9]) or the closely related one that exploits sequential decision-making based on drift-diffusion models (e.g. [10,31]). With respect to the analyses needed here such models are overly complex, given their aim of actually simulating gaze shifts, and/or just suitable to cope with simple visual stimuli.…”

Using Gaze for Behavioural Biometrics

“…Notably, such view allows to treat fixations and saccades as realisations of processes exhibiting the same dynamics, namely an O-U process. Following [ 95 , 96 ], the sequence of fixations and saccades composing a scan path can be described via a state-dependent SDE defining a 2D process.…”

“…Denote the state variable indicating whether at time t either a fixation or a saccade is performed [ 95 , 96 ]; in other words, is a switching binary RV accounting for the transition from the current fixation to the next saccade and vice versa. Then, Equation ( 3 ), can be written in the state-dependent form: where and noise are now the state-dependent parameters and noise process, respectively; in simple terms, depending on , Equation ( 5 ) can describe either local fixational dynamics or global, saccadic displacements.…”

“…Recently, theoretical models have been proposed to simulate gaze behaviour based on composite O-U processes in the framework of either a stochastic version of the Marginal Value Theorem [ 95 ] or of multi-alternative perceptual decision making [ 96 ]. However, the advantage of having general solutions derived from first principles is mitigated by both the complexity in determining the parameters and the computational cost that are likely to burden biometric applications.…”