Hippocampal representations as a function of time, subregion, and brain state

https://doi.org/10.1016/j.nlm.2018.03.006Get rights and content

Highlights

  • Hippocampal representations highlight shared or distinctive features of experiences.

  • Preferred representations vary by subfield, neuronal excitability and neuromodulation.

  • DG separates to support memory precision, while CA1 integrates for flexibility.

  • Sustained neuronal excitability may link related experiences across time.

  • Salience triggers release of neuromodulators, promoting separation or integration.

Abstract

How does the hippocampus represent interrelated experiences in memory? We review prominent yet seemingly contradictory theoretical perspectives, which propose that the hippocampus distorts experiential representations to either emphasize their distinctiveness or highlight common elements. These fundamentally different kinds of memory representations may be instantiated in the brain via conjunctive separated codes and adaptively differentiated codes on the one hand, or integrated relational codes on the other. After reviewing empirical support for these different coding schemes within the hippocampus, we outline two organizing principles which may explain the conflicting findings in the literature. First focusing on where the memories are formed and stored, we argue that distinct hippocampal regions represent experiences at multiple levels of abstraction and may transmit them to distinct cortical networks. Then focusing on when memories are formed, we identify several factors that can open and maintain specialized time windows, during which the very same hippocampal network is biased toward one coding scheme over the others. Specifically, we discuss evidence for (1) excitability-mediated integration windows, maintained by persistently elevated CREB levels following encoding of a specific memory, (2) fleeting cholinergically-mediated windows favoring memory separation, and (3) sustained dopaminergically-mediated windows favoring memory integration. By presenting a broad overview of different hippocampal coding schemes across species, we hope to inspire future empirical and modeling research to consider how factors surrounding memory formation shape the representations in which they are stored.

Introduction

It has been over half a century since patient HM’s surgery first demonstrated the critical role of the hippocampus in memory formation (Scoville & Milner, 1957), and seemingly countless researchers are still enthralled by this complex, mysterious structure. Indeed, we learn about new and important hippocampal functions every day; in addition to memory formation and retrieval, new findings suggest its role in flexible cognition more broadly, including navigation (Astur et al., 2002, Hirshhorn et al., 2012), reasoning (Sheldon et al., 2011, Zeithamova et al., 2012), decision making (Barron et al., 2013, Palombo et al., 2015, Shohamy and Daw, 2015), and imagination (Addis and Schacter, 2011, Mullally and Maguire, 2014)—to name just a few. How does the hippocampus support such diverse behaviours? One possible answer is in its apparent ability to represent our experiences at multiple levels of abstraction, allowing our memories to be simultaneously high fidelity and flexible.

Theoretical and computational modeling of these hippocampal representational schemes have guided memory research since the 1970’s (Marr, 1971). These ideas, however, have only recently become the subject of empirical investigation, in part due to prior barriers in ‘reading out’ the structure of hippocampal representations. The recent wellspring of machine learning and pattern analysis approaches in both human (Kriegeskorte, Mur, & Bandettini, 2008) and rodent (McKenzie et al., 2016) neuroscience combined with advances in data acquisition methods finally enable the characterization of memory-specific neural patterns required to test these entrenched theories (Box 1). In this review, we provide a high-level overview of different theoretical representational schemes along with recent evidence from rodent and human literatures. Throughout, we focus on how hippocampal contributions to memory are shaped by the nature of its representations and memory traces, terms we use to refer to any pattern of neural activity that encodes a specific memory and enables subsequent retrieval.

As we review below, however, the answers emerging from representational investigations are not always straightforward. Rather than supporting the simple dominance of one scheme over the others, data suggest that the hippocampus may employ multiple schemes. Without organizing principles or the identification of key regulatory factors, this could be a theoretical calamity for memory research. Here, we focus on two factors that determine the representational code employed within a hippocampal network: (1) location within the hippocampus, as different subregions are biased toward different kinds of representations and may have separate outputs; (2) the state of neurons leading up to a given experience, including their excitability levels and the surrounding concentrations of specific neuromodulators.

Section snippets

Making memories for related events

In the lab, we often engineer to-be-remembered events to be as distinct as possible. However, such distinctiveness is supremely artificial. In the real world, so many of our experiences are interconnected via familiar people, places, and things. How do we store memories for such interrelated events? This review will focus on three prominent types of codes—pattern separation, integration, and differentiation—which are defined by the representational transformations they perform, alternatively

The emergence of complementary codes across hippocampal subregions

How might the hippocampus simultaneously maintain these different kinds of representations? One answer may lie in the heterogeneity of hippocampal structure and function across the transverse, longitudinal and radial planes. Here, we focus on heterogeneity in the transverse plane, reviewing both classic models and recent evidence for representational differences across subfields, beginning in the dentate gyrus (DG) then moving through the cornu ammonis fields (CA) 3 and 1. See Box 3 for

Time windows for memory integration and separation

So far, we have treated hippocampal neurons and networks as though they always operate in the same state—i.e., the same two related experiences will always be integrated or separated, with the coding scheme automatically determined by the fixed wiring properties of the hippocampus. Here, we will introduce an essential layer of complexity by considering dynamic molecular mechanisms. Specifically, we propose that shifting concentrations of key molecules open what we call integration or separation

Conclusions

A major theoretical challenge is facing memory researchers today—we do not have a clear model of how the hippocampus represents information in memory. At first blush, this statement may seem shocking. How could such a fundamental property of one of the most investigated brain regions remain unknown? Moreover, longstanding models are so entrenched in our work that they can give the false sense of consensus. Some researchers may believe that the hippocampus performs pattern separation and

Acknowledgements

This work was supported by NSERC Discovery Grant 500491 and CFI/ORF Project #34479 to KD. The authors declare no competing interests.

References (210)

  • L. Descarries et al.

    Diffuse transmission by acetylcholine in the CNS

    Progress in Neurobiology

    (1997)
  • G.J. Detre et al.

    Moderate levels of activation lead to forgetting in the think/no-think paradigm

    Neuropsychologia

    (2013)
  • A. Easton et al.

    A specific role for septohippocampal acetylcholine in memory?

    Neuropsychologia

    (2012)
  • H. Eichenbaum

    Hippocampus

    Neuron

    (2004)
  • H. Eichenbaum et al.

    The hippocampus, memory, and place cells: is it spatial memory or a memory space?

    Neuron

    (1999)
  • J.A. Etzel et al.

    Searchlight analysis: Promise, pitfalls, and potential

    NeuroImage

    (2013)
  • M.S. Fanselow et al.

    Are the dorsal and ventral hippocampus functionally distinct structures?

    Neuron

    (2010)
  • A. Gasbarri et al.

    Anterograde and retrograde tracing of projections from the ventral tegmental area to the hippocampal formation in the rat

    Brain Research Bulletin

    (1994)
  • M.J. Gruber et al.

    Post-learning hippocampal dynamics promote preferential retention of rewarding events

    Neuron

    (2016)
  • M.J. Gruber et al.

    Expected reward modulates encoding-related theta activity before an event

    NeuroImage

    (2013)
  • J. Guzowski et al.

    Ensemble dynamics of hippocampal regions CA3 and CA1

    Neuron

    (2004)
  • Hasselmo

    Neuromodulation: acetylcholine and memory consolidation

    Trends in Cognitive Sciences

    (1999)
  • M.E. Hasselmo et al.

    High acetylcholine levels set circuit dynamics for attention and encoding and low acetylcholine levels set dynamics for consolidation

    Program of Brain Research

    (2004)
  • J.D. Haynes

    A primer on pattern-based approaches to fmri: principles, pitfalls, and perspectives

    Neuron

    (2015)
  • P.T. Huerta et al.

    Bidirectional synaptic plasticity induced by a single burst during cholinergic theta oscillation in CA1 in vitro

    Neuron

    (1995)
  • R.A. Hut et al.

    The cholinergic system, circadian rhythmicity, and time memory

    Behavioural Brain Research

    (2011)
  • I. Klinkenberg et al.

    Acetylcholine and attention

    Behavioural Brain Research

    (2011)
  • J.J. Knierim et al.

    Tracking the flow of hippocampal computation: Pattern separation, pattern completion, and attractor dynamics

    Neurobiology of Learning and Memory

    (2016)
  • N. Kriegeskorte et al.

    Matching categorical object representations in inferior temporal cortex of man and monkey

    Neuron

    (2008)
  • D.R. Addis et al.

    The hippocampus and imagining the future: Where do we stand?

    Frontiers in Human Neuroscience

    (2011)
  • M. Aly et al.

    Attention promotes episodic encoding by stabilizing hippocampal representations

    Proceedings of the National Academy of Sciences

    (2016)
  • M. Aly et al.

    How Hippocampal Memory Shapes, and Is Shaped by, Attention

  • Amaral, D. G., & Lavenex, P. (2006). Hippocampal neuroanatomy. (D. G. Amaral, P. Andersen, T. V Bliss, R. G. Morris, &...
  • M.C. Anderson et al.

    Integration as a general boundary condition on retrieval-induced forgetting

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (1999)
  • A. Atri et al.

    Blockade of central cholinergic receptors impairs new learning and increases proactive interference in a word paired-associate memory task

    Behavioral Neuroscience

    (2004)
  • M. Azab et al.

    Contributions of human hippocampal subfields to spatial and temporal pattern separation

    Hippocampus

    (2014)
  • A. Bakker et al.

    Pattern separation in the human hippocampal CA3 and dentate gyrus

    Science (New York, N.Y.)

    (2008)
  • H.C. Barron et al.

    Online evaluation of novel choices by simultaneous representation of multiple memories

    Nature Neuroscience

    (2013)
  • H.C. Barron et al.

    Repetition suppression: A means to index neural representations using BOLD?

    Philosophical Transactions of the Royal Society B: Biological Sciences

    (2016)
  • D. Berron et al.

    Strong evidence for pattern separation in human dentate gyrus

    Journal of Neuroscience

    (2016)
  • I. Bethus et al.

    Dopamine and memory: modulation of the persistence of memory for novel hippocampal NMDA receptor-dependent paired associates

    Journal of Neuroscience

    (2010)
  • Bostock, E., Muller, R. U., & Kubie-i-, J. L. (1991). Experience-dependent modifications of hippocampal place cell...
  • T.H. Brown et al.

    Voltage-clamp analysis of mossy fiber synaptic input to hippocampal neurons

    Journal of Neurophysiology

    (1983)
  • D.J. Cai et al.

    A shared neural ensemble links distinct contextual memories encoded close in time

    Nature

    (2016)
  • G. Carlson et al.

    Endocannabinoids facilitate the induction of LTP in the hippocampus

    Nature Neuroscience

    (2002)
  • M.F. Carr et al.

    Hippocampal replay in the awake state: A potential substrate for memory consolidation and retrieval

    Nature Neuroscience

    (2011)
  • A.J.H. Chanales et al.

    Overlap among spatial memories triggers repulsion of hippocampal representations article overlap among spatial memories triggers repulsion of hippocampal representations

    Current Biology

    (2017)
  • J. Chen et al.

    Associative retrieval processes in the human medial temporal lobe: hippocampal retrieval success and CA1 mismatch detection

    Learning & Memory (Cold Spring Harbor, N.Y.)

    (2011)
  • R.M. Cichy et al.

    Similarity-based fusion of MEG and fMRI reveals spatio-temporal dynamics in human cortex during visual object recognition

    Cerebral Cortex

    (2016)
  • B.J. Claiborne et al.

    A light and electron microscopic analysis of the mossy fibers of the rat dentate gyrus

    The Journal of Comparative Neurology

    (1986)
  • Cited by (49)

    • Exploration patterns shape cognitive map learning

      2023, Cognition
      Citation Excerpt :

      For example, when information is organized as a structured graph, humans tend to pick up on clusters that are closest together on relevant dimension, such as time, space, or semantic distance. Establishing these clusters, or communities, enables us to represent tasks hierarchically and flexibly (Karuza et al., 2016; Solway et al., 2014) as we build structured knowledge by accumulating pieces of information over time (Duncan & Schlichting, 2018; Schapiro, Rogers, Cordova, Turk-Browne, & Botvinick, 2013; Schapiro, Turk-Browne, Norman, & Botvinick, 2016). Simulations and human behavioral experiments suggest that the structure of such information graphs is most useful to us if the communities are tightly clustered into ‘neighborhoods’ of information, highlighting that environments or knowledge structures with different levels of complexity also have different levels of learnability (Karuza et al., 2016; Lynn et al., 2020).

    • Stress disrupts insight-driven mnemonic reconfiguration in the medial temporal lobe

      2023, NeuroImage
      Citation Excerpt :

      Interestingly, we found that while control participants exhibited a significant increase in representational dissimilarity for linked events from pre to post insight in the right anterior hippocampus, stress abolished this insight-related change in anterior hippocampal representations (group × time × link interaction: F(1, 53) = 6.20, pcorr = .032, ηG = .017; Fig. 7C). For the posterior hippocampus, there was no such change (group x time × link interaction: F(1, 56) = 1.03, pcorr = .626, ηG = .002; Fig. S3), in line with previous studies suggesting that the anterior but not the posterior part of the hippocampus is involved in mnemonic integration (Collin et al., 2015; Dandolo and Schwabe, 2018; De Shetler and Rissman, 2017; Duncan and Schlichting, 2018; Morton et al., 2017; Robin and Moscovitch, 2017). We performed a follow-up analysis of the interaction in the anterior hippocampus and found that controls showed a significant increase in representational dissimilarity from pre to post specifically for linked events (t(26) = -2.13, p = .043, drepeated measures = .41; Fig. 7C) but no increase in representational dissimilarity from pre to post for non-linked events (t(26) = .05, p = .620, drepeated measures = -.10; time × link interaction: F(1, 26) = 4.51, p = .043, ηG = .027).

    View all citing articles on Scopus
    1

    Equal contributions.

    View full text