An alternating renewal process describes the buildup of perceptual segregation

Sara A. Steele, Daniel Tranchina, John Rinze

Research output: Contribution to journalArticle

Abstract

For some ambiguous scenes perceptual conflict arises between integration and segregation. Initially, all stimulus features seem integrated. Then abruptly, perhaps after a few seconds, a segregated percept emerges. For example, segregation of acoustic features into streams may require several seconds. In behavioral experiments, when a subject's reports of stream segregation are averaged over repeated trials, one obtains a buildup function, a smooth time course for segregation probability. The buildup function has been said to reflect an underlying mechanism of evidence accumulation or adaptation. During long duration stimuli perception may alternate between integration and segregation. We present a statistical model based on an alternating renewal process (ARP) that generates buildup functions without an accumulative process. In our model, perception alternates during a trial between different groupings, as in perceptual bistability, with random and independent dominance durations sampled from different percept-specific probability distributions. Using this theory, we describe the short-term dynamics of buildup observed on short trials in terms of the long-term statistics of percept durations for the two alternating perceptual organizations. Our statistical-dynamics model describes well the buildup functions and alternations in simulations of pseudo-mechanistic neuronal network models with percept-selective populations competing through mutual inhibition. Even though the competition model can show history dependence through slow adaptation, our statistical switching model, that neglects history, predicts well the buildup function. We propose that accumulation is not a necessary feature to produce buildup. Generally, if alternations between two states exhibit independent durations with stationary statistics then the associated buildup function can be described by the statistical dynamics of an ARP.

Original languageEnglish (US)
Article number166
Pages (from-to)1-13
Number of pages13
JournalFrontiers in Computational Neuroscience
Volume8
Issue numberJAN
DOIs
StatePublished - 2015

Fingerprint

Statistical Models
History
Acoustics
Population

Keywords

  • Alternating renewal process
  • Bistable perception
  • Buildup
  • Perceptual dynamics
  • Perceptual organization
  • Stream segregation

ASJC Scopus subject areas

  • Neuroscience (miscellaneous)
  • Cellular and Molecular Neuroscience

Cite this

An alternating renewal process describes the buildup of perceptual segregation. / Steele, Sara A.; Tranchina, Daniel; Rinze, John.

In: Frontiers in Computational Neuroscience, Vol. 8, No. JAN, 166, 2015, p. 1-13.

Research output: Contribution to journalArticle

@article{f7dd5b02b3554e51b3f21bdad035ec0d,
title = "An alternating renewal process describes the buildup of perceptual segregation",
abstract = "For some ambiguous scenes perceptual conflict arises between integration and segregation. Initially, all stimulus features seem integrated. Then abruptly, perhaps after a few seconds, a segregated percept emerges. For example, segregation of acoustic features into streams may require several seconds. In behavioral experiments, when a subject's reports of stream segregation are averaged over repeated trials, one obtains a buildup function, a smooth time course for segregation probability. The buildup function has been said to reflect an underlying mechanism of evidence accumulation or adaptation. During long duration stimuli perception may alternate between integration and segregation. We present a statistical model based on an alternating renewal process (ARP) that generates buildup functions without an accumulative process. In our model, perception alternates during a trial between different groupings, as in perceptual bistability, with random and independent dominance durations sampled from different percept-specific probability distributions. Using this theory, we describe the short-term dynamics of buildup observed on short trials in terms of the long-term statistics of percept durations for the two alternating perceptual organizations. Our statistical-dynamics model describes well the buildup functions and alternations in simulations of pseudo-mechanistic neuronal network models with percept-selective populations competing through mutual inhibition. Even though the competition model can show history dependence through slow adaptation, our statistical switching model, that neglects history, predicts well the buildup function. We propose that accumulation is not a necessary feature to produce buildup. Generally, if alternations between two states exhibit independent durations with stationary statistics then the associated buildup function can be described by the statistical dynamics of an ARP.",
keywords = "Alternating renewal process, Bistable perception, Buildup, Perceptual dynamics, Perceptual organization, Stream segregation",
author = "Steele, {Sara A.} and Daniel Tranchina and John Rinze",
year = "2015",
doi = "10.3389/fncom.2014.00166",
language = "English (US)",
volume = "8",
pages = "1--13",
journal = "Frontiers in Computational Neuroscience",
issn = "1662-5188",
publisher = "Frontiers Research Foundation",
number = "JAN",

}

TY - JOUR

T1 - An alternating renewal process describes the buildup of perceptual segregation

AU - Steele, Sara A.

AU - Tranchina, Daniel

AU - Rinze, John

PY - 2015

Y1 - 2015

N2 - For some ambiguous scenes perceptual conflict arises between integration and segregation. Initially, all stimulus features seem integrated. Then abruptly, perhaps after a few seconds, a segregated percept emerges. For example, segregation of acoustic features into streams may require several seconds. In behavioral experiments, when a subject's reports of stream segregation are averaged over repeated trials, one obtains a buildup function, a smooth time course for segregation probability. The buildup function has been said to reflect an underlying mechanism of evidence accumulation or adaptation. During long duration stimuli perception may alternate between integration and segregation. We present a statistical model based on an alternating renewal process (ARP) that generates buildup functions without an accumulative process. In our model, perception alternates during a trial between different groupings, as in perceptual bistability, with random and independent dominance durations sampled from different percept-specific probability distributions. Using this theory, we describe the short-term dynamics of buildup observed on short trials in terms of the long-term statistics of percept durations for the two alternating perceptual organizations. Our statistical-dynamics model describes well the buildup functions and alternations in simulations of pseudo-mechanistic neuronal network models with percept-selective populations competing through mutual inhibition. Even though the competition model can show history dependence through slow adaptation, our statistical switching model, that neglects history, predicts well the buildup function. We propose that accumulation is not a necessary feature to produce buildup. Generally, if alternations between two states exhibit independent durations with stationary statistics then the associated buildup function can be described by the statistical dynamics of an ARP.

AB - For some ambiguous scenes perceptual conflict arises between integration and segregation. Initially, all stimulus features seem integrated. Then abruptly, perhaps after a few seconds, a segregated percept emerges. For example, segregation of acoustic features into streams may require several seconds. In behavioral experiments, when a subject's reports of stream segregation are averaged over repeated trials, one obtains a buildup function, a smooth time course for segregation probability. The buildup function has been said to reflect an underlying mechanism of evidence accumulation or adaptation. During long duration stimuli perception may alternate between integration and segregation. We present a statistical model based on an alternating renewal process (ARP) that generates buildup functions without an accumulative process. In our model, perception alternates during a trial between different groupings, as in perceptual bistability, with random and independent dominance durations sampled from different percept-specific probability distributions. Using this theory, we describe the short-term dynamics of buildup observed on short trials in terms of the long-term statistics of percept durations for the two alternating perceptual organizations. Our statistical-dynamics model describes well the buildup functions and alternations in simulations of pseudo-mechanistic neuronal network models with percept-selective populations competing through mutual inhibition. Even though the competition model can show history dependence through slow adaptation, our statistical switching model, that neglects history, predicts well the buildup function. We propose that accumulation is not a necessary feature to produce buildup. Generally, if alternations between two states exhibit independent durations with stationary statistics then the associated buildup function can be described by the statistical dynamics of an ARP.

KW - Alternating renewal process

KW - Bistable perception

KW - Buildup

KW - Perceptual dynamics

KW - Perceptual organization

KW - Stream segregation

UR - http://www.scopus.com/inward/record.url?scp=84920564043&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84920564043&partnerID=8YFLogxK

U2 - 10.3389/fncom.2014.00166

DO - 10.3389/fncom.2014.00166

M3 - Article

VL - 8

SP - 1

EP - 13

JO - Frontiers in Computational Neuroscience

JF - Frontiers in Computational Neuroscience

SN - 1662-5188

IS - JAN

M1 - 166

ER -