Competition, gating and learning: A new computational model of task switching

Michael Todd, Kong-Fatt Wong, Jonathan Cohen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Task switching has been the subject of extensive research, as it offers a valuable window into the dynamics by which cognitive control states interact with each other and are updated in response to shifting contextual demands. Recent work suggests that gating, learning, and performance monitoring mechanisms play key, interrelated roles in the emergence of counterintuitive task switching phenomena such as residual switch costs (Rogers & Monsell, 1995), backward inhibition (Mayr & Keele, 2000), and asymmetrical switch costs (Wylie & Allport, 2000). We introduce a new computational model of these phenomena. Based on Guided Activation Theory (Miller & Cohen, 2001), the model adds gating (Braver & Cohen, 1999), performance monitoring (Botvinick et al., 2001), and learning mechanisms, combining the latter two into a novel Competition-Driven Learning (CDL) rule which is related to Norman et al.’s biologically plausible oscillatory learning rule (2006). CDL uses response conflict to drive inhibitory learning between incompatible task representations, punishing irrelevant, interfering tasks. The model successfully reproduces the above phenomena. Lesions of the model’s gating, learning, and performance monitoring components demonstrate that a simpler model cannot accommodate the same findings equally well. Finally, the model makes novel predictions about task switching, such as the shape of an entire time series of RT distributions, which would not have been testable without an explicit computational model.
LanguageEnglish
Title of host publicationUnknown Host Publication
Number of pages1
Publication statusPublished - 2007
EventSociety for Neuroscience 2007 - San Diego
Duration: 1 Jan 2007 → …

Conference

ConferenceSociety for Neuroscience 2007
Period1/01/07 → …

Fingerprint

performance monitoring
learning
government supervision
costs
activation
time series

Cite this

Todd, M., Wong, K-F., & Cohen, J. (2007). Competition, gating and learning: A new computational model of task switching. In Unknown Host Publication
Todd, Michael ; Wong, Kong-Fatt ; Cohen, Jonathan. / Competition, gating and learning: A new computational model of task switching. Unknown Host Publication. 2007.
@inproceedings{9a3be278b02e402ab753ded78294c678,
title = "Competition, gating and learning: A new computational model of task switching",
abstract = "Task switching has been the subject of extensive research, as it offers a valuable window into the dynamics by which cognitive control states interact with each other and are updated in response to shifting contextual demands. Recent work suggests that gating, learning, and performance monitoring mechanisms play key, interrelated roles in the emergence of counterintuitive task switching phenomena such as residual switch costs (Rogers & Monsell, 1995), backward inhibition (Mayr & Keele, 2000), and asymmetrical switch costs (Wylie & Allport, 2000). We introduce a new computational model of these phenomena. Based on Guided Activation Theory (Miller & Cohen, 2001), the model adds gating (Braver & Cohen, 1999), performance monitoring (Botvinick et al., 2001), and learning mechanisms, combining the latter two into a novel Competition-Driven Learning (CDL) rule which is related to Norman et al.’s biologically plausible oscillatory learning rule (2006). CDL uses response conflict to drive inhibitory learning between incompatible task representations, punishing irrelevant, interfering tasks. The model successfully reproduces the above phenomena. Lesions of the model’s gating, learning, and performance monitoring components demonstrate that a simpler model cannot accommodate the same findings equally well. Finally, the model makes novel predictions about task switching, such as the shape of an entire time series of RT distributions, which would not have been testable without an explicit computational model.",
author = "Michael Todd and Kong-Fatt Wong and Jonathan Cohen",
note = "Program#/Poster#: 634.10/CCC3 Reference text: Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict Monitoring and Cognitive Control. Psychological Review , 108 (3), 624-652. Braver, T. S., & Cohen, J. D. (1999). Dopamine, cognitive control, and schizophrenia: The gating model. In J. A. Reggia, E. Ruppin, & D. Glanzman (Eds.), Progress in Brain Research (pp. 327-349). Amsterdam, North-Holland: Elsevier Science. Mayr, U., & Keele, S. W. (2000). Changing Internal Constraints on Action: The Role of Backward Inhibition. Journal of Experimental Psychology: General , 129 (1), 4-26. Miller, E. K., & Cohen, J. D. (2001). An Interactive Theory of Prefrontal Cortex Function. Annual Review of Neuroscience , 24, 167-202. Norman, K. A., Newman, E., Detre, G., & Polyn, S. (2006). How Inhibitory Oscillations Can Train Neural Networks and Punish Competitors. Neural Computation , 18, 1577-1610. Rogers, R. D., & Monsell, S. (1995). Costs of a Predictable Switch Between Simple Cognitive Tasks. Journal of Experimental Psychology: General , 124 (2), 207-231. Wylie, G., & Allport, A. (2000). Task Switching and the Measurement of {"}Switch Costs{"}. Psychological Research , 63, 212-233.",
year = "2007",
language = "English",
booktitle = "Unknown Host Publication",

}

Todd, M, Wong, K-F & Cohen, J 2007, Competition, gating and learning: A new computational model of task switching. in Unknown Host Publication. Society for Neuroscience 2007, 1/01/07.

Competition, gating and learning: A new computational model of task switching. / Todd, Michael; Wong, Kong-Fatt; Cohen, Jonathan.

Unknown Host Publication. 2007.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Competition, gating and learning: A new computational model of task switching

AU - Todd, Michael

AU - Wong, Kong-Fatt

AU - Cohen, Jonathan

N1 - Program#/Poster#: 634.10/CCC3 Reference text: Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict Monitoring and Cognitive Control. Psychological Review , 108 (3), 624-652. Braver, T. S., & Cohen, J. D. (1999). Dopamine, cognitive control, and schizophrenia: The gating model. In J. A. Reggia, E. Ruppin, & D. Glanzman (Eds.), Progress in Brain Research (pp. 327-349). Amsterdam, North-Holland: Elsevier Science. Mayr, U., & Keele, S. W. (2000). Changing Internal Constraints on Action: The Role of Backward Inhibition. Journal of Experimental Psychology: General , 129 (1), 4-26. Miller, E. K., & Cohen, J. D. (2001). An Interactive Theory of Prefrontal Cortex Function. Annual Review of Neuroscience , 24, 167-202. Norman, K. A., Newman, E., Detre, G., & Polyn, S. (2006). How Inhibitory Oscillations Can Train Neural Networks and Punish Competitors. Neural Computation , 18, 1577-1610. Rogers, R. D., & Monsell, S. (1995). Costs of a Predictable Switch Between Simple Cognitive Tasks. Journal of Experimental Psychology: General , 124 (2), 207-231. Wylie, G., & Allport, A. (2000). Task Switching and the Measurement of "Switch Costs". Psychological Research , 63, 212-233.

PY - 2007

Y1 - 2007

N2 - Task switching has been the subject of extensive research, as it offers a valuable window into the dynamics by which cognitive control states interact with each other and are updated in response to shifting contextual demands. Recent work suggests that gating, learning, and performance monitoring mechanisms play key, interrelated roles in the emergence of counterintuitive task switching phenomena such as residual switch costs (Rogers & Monsell, 1995), backward inhibition (Mayr & Keele, 2000), and asymmetrical switch costs (Wylie & Allport, 2000). We introduce a new computational model of these phenomena. Based on Guided Activation Theory (Miller & Cohen, 2001), the model adds gating (Braver & Cohen, 1999), performance monitoring (Botvinick et al., 2001), and learning mechanisms, combining the latter two into a novel Competition-Driven Learning (CDL) rule which is related to Norman et al.’s biologically plausible oscillatory learning rule (2006). CDL uses response conflict to drive inhibitory learning between incompatible task representations, punishing irrelevant, interfering tasks. The model successfully reproduces the above phenomena. Lesions of the model’s gating, learning, and performance monitoring components demonstrate that a simpler model cannot accommodate the same findings equally well. Finally, the model makes novel predictions about task switching, such as the shape of an entire time series of RT distributions, which would not have been testable without an explicit computational model.

AB - Task switching has been the subject of extensive research, as it offers a valuable window into the dynamics by which cognitive control states interact with each other and are updated in response to shifting contextual demands. Recent work suggests that gating, learning, and performance monitoring mechanisms play key, interrelated roles in the emergence of counterintuitive task switching phenomena such as residual switch costs (Rogers & Monsell, 1995), backward inhibition (Mayr & Keele, 2000), and asymmetrical switch costs (Wylie & Allport, 2000). We introduce a new computational model of these phenomena. Based on Guided Activation Theory (Miller & Cohen, 2001), the model adds gating (Braver & Cohen, 1999), performance monitoring (Botvinick et al., 2001), and learning mechanisms, combining the latter two into a novel Competition-Driven Learning (CDL) rule which is related to Norman et al.’s biologically plausible oscillatory learning rule (2006). CDL uses response conflict to drive inhibitory learning between incompatible task representations, punishing irrelevant, interfering tasks. The model successfully reproduces the above phenomena. Lesions of the model’s gating, learning, and performance monitoring components demonstrate that a simpler model cannot accommodate the same findings equally well. Finally, the model makes novel predictions about task switching, such as the shape of an entire time series of RT distributions, which would not have been testable without an explicit computational model.

M3 - Conference contribution

BT - Unknown Host Publication

ER -