Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach

Ali Yousefi, Ishita Basu, Angelique C. Paulk, Noam Peled, Emad N. Eskandar, Darin D. Dougherty, Sydney S. Cash, Alik S. Widge, Uri T. Eden

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Cognitive processes, such as learning and cognitive flexibility, are both difficult to measure and to sample continuously using objective tools because cognitive processes arise from distributed, high-dimensional neural activity. For both research and clinical applications, that dimensionality must be reduced. To reduce dimensionality and measure underlying cognitive processes, we propose a modeling framework in which a cognitive process is defined as a low-dimensional dynamical latent variable-called a cognitive state, which links high-dimensional neural recordings and multidimensional behavioral readouts. This framework allows us to decompose the hard problem of modeling the relationship between neural and behavioral data into separable encoding-decoding approaches. We first use a state-space modeling framework, the behavioral decoder, to articulate the relationship between an objective behavioral readout (e.g., response times) and cognitive state. The second step, the neural encoder, involves using a generalized linear model (GLM) to identify the relationship between the cognitive state and neural signals, such as local field potential (LFP). We then use the neural encoder model and a Bayesian filter to estimate cognitive state using neural data (LFP power) to generate the neural decoder. We provide goodness-of-fit analysis and model selection criteria in support of the encoding-decoding result. We apply this framework to estimate an underlying cognitive state from neural data in human participants (formula presented ) performing a cognitive conflict task. We successfully estimated the cognitive state within the 95% confidence intervals of that estimated using behavior readout for an average of 90% of task trials across participants. In contrast to previous encoder-decoder models, our proposed modeling framework incorporates LFP spectral power to encode and decode a cognitive state. The framework allowed us to capture the temporal evolution of the underlying cognitive processes, which could be key to the development of closed-loop experiments and treatments.

Original languageEnglish (US)
Pages (from-to)1751-1788
Number of pages38
JournalNeural computation
Volume31
Issue number9
DOIs
StatePublished - Sep 1 2019

Fingerprint

Bayes Theorem
Patient Selection
Reaction Time
Linear Models
Learning
Confidence Intervals
Research
Power (Psychology)
Decoding
Physiology
Cognitive State
Therapeutics
Cognitive Processes
Modeling
Conflict (Psychology)

ASJC Scopus subject areas

  • Arts and Humanities (miscellaneous)
  • Cognitive Neuroscience

Cite this

Yousefi, A., Basu, I., Paulk, A. C., Peled, N., Eskandar, E. N., Dougherty, D. D., ... Eden, U. T. (2019). Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach. Neural computation, 31(9), 1751-1788. https://doi.org/10.1162/neco_a_01196

Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach. / Yousefi, Ali; Basu, Ishita; Paulk, Angelique C.; Peled, Noam; Eskandar, Emad N.; Dougherty, Darin D.; Cash, Sydney S.; Widge, Alik S.; Eden, Uri T.

In: Neural computation, Vol. 31, No. 9, 01.09.2019, p. 1751-1788.

Research output: Contribution to journalArticle

Yousefi, A, Basu, I, Paulk, AC, Peled, N, Eskandar, EN, Dougherty, DD, Cash, SS, Widge, AS & Eden, UT 2019, 'Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach', Neural computation, vol. 31, no. 9, pp. 1751-1788. https://doi.org/10.1162/neco_a_01196
Yousefi, Ali ; Basu, Ishita ; Paulk, Angelique C. ; Peled, Noam ; Eskandar, Emad N. ; Dougherty, Darin D. ; Cash, Sydney S. ; Widge, Alik S. ; Eden, Uri T. / Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach. In: Neural computation. 2019 ; Vol. 31, No. 9. pp. 1751-1788.
@article{b95eb235ebdc43fb899acd6a468c97c1,
title = "Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach",
abstract = "Cognitive processes, such as learning and cognitive flexibility, are both difficult to measure and to sample continuously using objective tools because cognitive processes arise from distributed, high-dimensional neural activity. For both research and clinical applications, that dimensionality must be reduced. To reduce dimensionality and measure underlying cognitive processes, we propose a modeling framework in which a cognitive process is defined as a low-dimensional dynamical latent variable-called a cognitive state, which links high-dimensional neural recordings and multidimensional behavioral readouts. This framework allows us to decompose the hard problem of modeling the relationship between neural and behavioral data into separable encoding-decoding approaches. We first use a state-space modeling framework, the behavioral decoder, to articulate the relationship between an objective behavioral readout (e.g., response times) and cognitive state. The second step, the neural encoder, involves using a generalized linear model (GLM) to identify the relationship between the cognitive state and neural signals, such as local field potential (LFP). We then use the neural encoder model and a Bayesian filter to estimate cognitive state using neural data (LFP power) to generate the neural decoder. We provide goodness-of-fit analysis and model selection criteria in support of the encoding-decoding result. We apply this framework to estimate an underlying cognitive state from neural data in human participants (formula presented ) performing a cognitive conflict task. We successfully estimated the cognitive state within the 95{\%} confidence intervals of that estimated using behavior readout for an average of 90{\%} of task trials across participants. In contrast to previous encoder-decoder models, our proposed modeling framework incorporates LFP spectral power to encode and decode a cognitive state. The framework allowed us to capture the temporal evolution of the underlying cognitive processes, which could be key to the development of closed-loop experiments and treatments.",
author = "Ali Yousefi and Ishita Basu and Paulk, {Angelique C.} and Noam Peled and Eskandar, {Emad N.} and Dougherty, {Darin D.} and Cash, {Sydney S.} and Widge, {Alik S.} and Eden, {Uri T.}",
year = "2019",
month = "9",
day = "1",
doi = "10.1162/neco_a_01196",
language = "English (US)",
volume = "31",
pages = "1751--1788",
journal = "Neural Computation",
issn = "0899-7667",
publisher = "MIT Press Journals",
number = "9",

}

TY - JOUR

T1 - Decoding Hidden Cognitive States From Behavior and Physiology Using a Bayesian Approach

AU - Yousefi, Ali

AU - Basu, Ishita

AU - Paulk, Angelique C.

AU - Peled, Noam

AU - Eskandar, Emad N.

AU - Dougherty, Darin D.

AU - Cash, Sydney S.

AU - Widge, Alik S.

AU - Eden, Uri T.

PY - 2019/9/1

Y1 - 2019/9/1

N2 - Cognitive processes, such as learning and cognitive flexibility, are both difficult to measure and to sample continuously using objective tools because cognitive processes arise from distributed, high-dimensional neural activity. For both research and clinical applications, that dimensionality must be reduced. To reduce dimensionality and measure underlying cognitive processes, we propose a modeling framework in which a cognitive process is defined as a low-dimensional dynamical latent variable-called a cognitive state, which links high-dimensional neural recordings and multidimensional behavioral readouts. This framework allows us to decompose the hard problem of modeling the relationship between neural and behavioral data into separable encoding-decoding approaches. We first use a state-space modeling framework, the behavioral decoder, to articulate the relationship between an objective behavioral readout (e.g., response times) and cognitive state. The second step, the neural encoder, involves using a generalized linear model (GLM) to identify the relationship between the cognitive state and neural signals, such as local field potential (LFP). We then use the neural encoder model and a Bayesian filter to estimate cognitive state using neural data (LFP power) to generate the neural decoder. We provide goodness-of-fit analysis and model selection criteria in support of the encoding-decoding result. We apply this framework to estimate an underlying cognitive state from neural data in human participants (formula presented ) performing a cognitive conflict task. We successfully estimated the cognitive state within the 95% confidence intervals of that estimated using behavior readout for an average of 90% of task trials across participants. In contrast to previous encoder-decoder models, our proposed modeling framework incorporates LFP spectral power to encode and decode a cognitive state. The framework allowed us to capture the temporal evolution of the underlying cognitive processes, which could be key to the development of closed-loop experiments and treatments.

AB - Cognitive processes, such as learning and cognitive flexibility, are both difficult to measure and to sample continuously using objective tools because cognitive processes arise from distributed, high-dimensional neural activity. For both research and clinical applications, that dimensionality must be reduced. To reduce dimensionality and measure underlying cognitive processes, we propose a modeling framework in which a cognitive process is defined as a low-dimensional dynamical latent variable-called a cognitive state, which links high-dimensional neural recordings and multidimensional behavioral readouts. This framework allows us to decompose the hard problem of modeling the relationship between neural and behavioral data into separable encoding-decoding approaches. We first use a state-space modeling framework, the behavioral decoder, to articulate the relationship between an objective behavioral readout (e.g., response times) and cognitive state. The second step, the neural encoder, involves using a generalized linear model (GLM) to identify the relationship between the cognitive state and neural signals, such as local field potential (LFP). We then use the neural encoder model and a Bayesian filter to estimate cognitive state using neural data (LFP power) to generate the neural decoder. We provide goodness-of-fit analysis and model selection criteria in support of the encoding-decoding result. We apply this framework to estimate an underlying cognitive state from neural data in human participants (formula presented ) performing a cognitive conflict task. We successfully estimated the cognitive state within the 95% confidence intervals of that estimated using behavior readout for an average of 90% of task trials across participants. In contrast to previous encoder-decoder models, our proposed modeling framework incorporates LFP spectral power to encode and decode a cognitive state. The framework allowed us to capture the temporal evolution of the underlying cognitive processes, which could be key to the development of closed-loop experiments and treatments.

UR - http://www.scopus.com/inward/record.url?scp=85071355489&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071355489&partnerID=8YFLogxK

U2 - 10.1162/neco_a_01196

DO - 10.1162/neco_a_01196

M3 - Article

C2 - 31335292

AN - SCOPUS:85071355489

VL - 31

SP - 1751

EP - 1788

JO - Neural Computation

JF - Neural Computation

SN - 0899-7667

IS - 9

ER -