A test of multiple correlation temporal window characteristic of non-Markov processes

TL;DR

This paper introduces a sensitive test for memory effects in successive events, revealing a temporal window where correlations exceed Markovian expectations, with evidence from cognitive visual tasks showing a peak around 2 seconds.

Contribution

The study presents a novel correlation-based test for detecting non-Markovian memory effects in sequential events, applied to cognitive perception data.

Findings

Identification of a K > 1 temporal window in visual perception tasks

Maximum correlation observed at approximately 2 seconds interval

Persistence of non-Markovian effects within a 1-second window around the peak

Abstract

We introduce a sensitive test of memory effects in successive events. The test consists of a combination K of binary correlations at successive times. K decays monotonically from K = 1 for uncorrelated events as a Markov process; whereas memory effects provide a temporal window with K > 1. For a monotonic memory fading, K < 1 always. Here we report evidence of a K > 1 temporal window in cognitive tasks consisting of the visual identification of the front face of the Necker cube after a previous presentation of the same. The K > 1 behaviour is maximal at an inter-measurement time {\tau} around 2 sec with inter-subject differences. The K > 1 persists over a time window of 1 sec around {\tau}; outside this window the K < 1 behaviour is recovered. The universal occurrence of a K > 1 window in pairs of successive perceptions suggests that, at variance with single visual stimuli eliciting a suitable response, a pair of stimuli shortly separated in time displays mutual correlations.