Hippocampal pattern separation supports reinforcement learning

Ian C. Ballard, Anthony D. Wagner, Samuel McClure

    Research output: Contribution to journalArticle

    Abstract

    Animals rely on learned associations to make decisions. Associations can be based on relationships between object features (e.g., the three leaflets of poison ivy leaves) and outcomes (e.g., rash). More often, outcomes are linked to multidimensional states (e.g., poison ivy is green in summer but red in spring). Feature-based reinforcement learning fails when the values of individual features depend on the other features present. One solution is to assign value to multi-featural conjunctive representations. Here, we test if the hippocampus forms separable conjunctive representations that enables the learning of response contingencies for stimuli of the form: AB+, B−, AC−, C+. Pattern analyses on functional MRI data show the hippocampus forms conjunctive representations that are dissociable from feature components and that these representations, along with those of cortex, influence striatal prediction errors. Our results establish a novel role for hippocampal pattern separation and conjunctive representation in reinforcement learning.

    Original languageEnglish (US)
    Article number1073
    JournalNature communications
    Volume10
    Issue number1
    DOIs
    StatePublished - Dec 1 2019

    Fingerprint

    Poisons
    Reinforcement learning
    Toxicodendron
    reinforcement
    learning
    Learning
    Hippocampus
    hippocampus
    poisons
    Corpus Striatum
    Animals
    Exanthema
    contingency
    Magnetic Resonance Imaging
    cortexes
    stimuli
    leaves
    summer
    animals
    alternating current

    ASJC Scopus subject areas

    • Chemistry(all)
    • Biochemistry, Genetics and Molecular Biology(all)
    • Physics and Astronomy(all)

    Cite this

    Hippocampal pattern separation supports reinforcement learning. / Ballard, Ian C.; Wagner, Anthony D.; McClure, Samuel.

    In: Nature communications, Vol. 10, No. 1, 1073, 01.12.2019.

    Research output: Contribution to journalArticle

    Ballard, Ian C. ; Wagner, Anthony D. ; McClure, Samuel. / Hippocampal pattern separation supports reinforcement learning. In: Nature communications. 2019 ; Vol. 10, No. 1.
    @article{d0984a5c9b7e4043acc49b21c9325a50,
    title = "Hippocampal pattern separation supports reinforcement learning",
    abstract = "Animals rely on learned associations to make decisions. Associations can be based on relationships between object features (e.g., the three leaflets of poison ivy leaves) and outcomes (e.g., rash). More often, outcomes are linked to multidimensional states (e.g., poison ivy is green in summer but red in spring). Feature-based reinforcement learning fails when the values of individual features depend on the other features present. One solution is to assign value to multi-featural conjunctive representations. Here, we test if the hippocampus forms separable conjunctive representations that enables the learning of response contingencies for stimuli of the form: AB+, B−, AC−, C+. Pattern analyses on functional MRI data show the hippocampus forms conjunctive representations that are dissociable from feature components and that these representations, along with those of cortex, influence striatal prediction errors. Our results establish a novel role for hippocampal pattern separation and conjunctive representation in reinforcement learning.",
    author = "Ballard, {Ian C.} and Wagner, {Anthony D.} and Samuel McClure",
    year = "2019",
    month = "12",
    day = "1",
    doi = "10.1038/s41467-019-08998-1",
    language = "English (US)",
    volume = "10",
    journal = "Nature Communications",
    issn = "2041-1723",
    publisher = "Nature Publishing Group",
    number = "1",

    }

    TY - JOUR

    T1 - Hippocampal pattern separation supports reinforcement learning

    AU - Ballard, Ian C.

    AU - Wagner, Anthony D.

    AU - McClure, Samuel

    PY - 2019/12/1

    Y1 - 2019/12/1

    N2 - Animals rely on learned associations to make decisions. Associations can be based on relationships between object features (e.g., the three leaflets of poison ivy leaves) and outcomes (e.g., rash). More often, outcomes are linked to multidimensional states (e.g., poison ivy is green in summer but red in spring). Feature-based reinforcement learning fails when the values of individual features depend on the other features present. One solution is to assign value to multi-featural conjunctive representations. Here, we test if the hippocampus forms separable conjunctive representations that enables the learning of response contingencies for stimuli of the form: AB+, B−, AC−, C+. Pattern analyses on functional MRI data show the hippocampus forms conjunctive representations that are dissociable from feature components and that these representations, along with those of cortex, influence striatal prediction errors. Our results establish a novel role for hippocampal pattern separation and conjunctive representation in reinforcement learning.

    AB - Animals rely on learned associations to make decisions. Associations can be based on relationships between object features (e.g., the three leaflets of poison ivy leaves) and outcomes (e.g., rash). More often, outcomes are linked to multidimensional states (e.g., poison ivy is green in summer but red in spring). Feature-based reinforcement learning fails when the values of individual features depend on the other features present. One solution is to assign value to multi-featural conjunctive representations. Here, we test if the hippocampus forms separable conjunctive representations that enables the learning of response contingencies for stimuli of the form: AB+, B−, AC−, C+. Pattern analyses on functional MRI data show the hippocampus forms conjunctive representations that are dissociable from feature components and that these representations, along with those of cortex, influence striatal prediction errors. Our results establish a novel role for hippocampal pattern separation and conjunctive representation in reinforcement learning.

    UR - http://www.scopus.com/inward/record.url?scp=85062625196&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85062625196&partnerID=8YFLogxK

    U2 - 10.1038/s41467-019-08998-1

    DO - 10.1038/s41467-019-08998-1

    M3 - Article

    VL - 10

    JO - Nature Communications

    JF - Nature Communications

    SN - 2041-1723

    IS - 1

    M1 - 1073

    ER -