Rate-distortion theory of neural coding and its implications for working memory

  1. Anthony MV Jakob  Is a corresponding author
  2. Samuel J Gershman
  1. Harvard University, United States


Rate-distortion theory provides a powerful framework for understanding the nature of human memory by formalizing the relationship between information rate (the average number of bits per stimulus transmitted across the memory channel) and distortion (the cost of memory errors). Here we show how this abstract computational-level framework can be realized by a model of neural population coding. The model reproduces key regularities of visual working memory, including some that were not previously explained by population coding models. We verify a novel prediction of the model by reanalyzing recordings of monkey prefrontal neurons during an oculomotor delayed response task.

Data availability

The current manuscript is a computational study, so no data have been generated for this manuscript. Source code can be found at https://github.com/amvjakob/wm-rate-distortion. The previously published datasets are available upon request from the corresponding authors of the published papers, Souza and Oberauer (2015), Daniel Bliss at al. (2017), Panichello et al. (2019). A minimally processed dataset from Barbosa et al. (2020) is available online ((https://github.com/comptelab/interplayPFC), with the raw data available upon request from the corresponding author of the published paper (raw monkey data available upon request to Christos Constantinidis cconstan@wakehealth.edu, and raw EEG data available upon request to Heike Stein, heike.c.stein@gmail.com). There are no specific application or approval processes involved in requesting these datasets.

The following previously published data sets were used

Article and author information

Author details

  1. Anthony MV Jakob

    Department of Neurobiology, Harvard University, Cambridge, United States
    For correspondence
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0996-1356
  2. Samuel J Gershman

    Department of Psychology, Harvard University, Cambridge, United States
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6546-3298


Fondation Bertarelli (Bertarelli Fellowship)

  • Anthony MV Jakob

National Science Foundation (NSF STC award,CCF-1231216)

  • Samuel J Gershman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Reviewing Editor

  1. Paul Bays, University of Cambridge, United Kingdom

Version history

  1. Preprint posted: March 2, 2022 (view preprint)
  2. Received: April 13, 2022
  3. Accepted: June 22, 2023
  4. Accepted Manuscript published: July 12, 2023 (version 1)
  5. Version of Record published: July 18, 2023 (version 2)


© 2023, Jakob & Gershman

This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.


  • 1,167
  • 240
  • 2

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Anthony MV Jakob
  2. Samuel J Gershman
Rate-distortion theory of neural coding and its implications for working memory
eLife 12:e79450.

Share this article


Further reading

    1. Genetics and Genomics
    2. Neuroscience
    Bohan Zhu, Richard I Ainsworth ... Javier González-Maeso
    Research Article

    Genome-wide association studies have revealed >270 loci associated with schizophrenia risk, yet these genetic factors do not seem to be sufficient to fully explain the molecular determinants behind this psychiatric condition. Epigenetic marks such as post-translational histone modifications remain largely plastic during development and adulthood, allowing a dynamic impact of environmental factors, including antipsychotic medications, on access to genes and regulatory elements. However, few studies so far have profiled cell-specific genome-wide histone modifications in postmortem brain samples from schizophrenia subjects, or the effect of antipsychotic treatment on such epigenetic marks. Here, we conducted ChIP-seq analyses focusing on histone marks indicative of active enhancers (H3K27ac) and active promoters (H3K4me3), alongside RNA-seq, using frontal cortex samples from antipsychotic-free (AF) and antipsychotic-treated (AT) individuals with schizophrenia, as well as individually matched controls (n=58). Schizophrenia subjects exhibited thousands of neuronal and non-neuronal epigenetic differences at regions that included several susceptibility genetic loci, such as NRG1, DISC1, and DRD3. By analyzing the AF and AT cohorts separately, we identified schizophrenia-associated alterations in specific transcription factors, their regulatees, and epigenomic and transcriptomic features that were reversed by antipsychotic treatment; as well as those that represented a consequence of antipsychotic medication rather than a hallmark of schizophrenia in postmortem human brain samples. Notably, we also found that the effect of age on epigenomic landscapes was more pronounced in frontal cortex of AT-schizophrenics, as compared to AF-schizophrenics and controls. Together, these data provide important evidence of epigenetic alterations in the frontal cortex of individuals with schizophrenia, and remark for the first time on the impact of age and antipsychotic treatment on chromatin organization.

    1. Neuroscience
    Aedan Yue Li, Natalia Ladyka-Wojcik ... Morgan Barense
    Research Article

    Combining information from multiple senses is essential to object recognition, core to the ability to learn concepts, make new inferences, and generalize across distinct entities. Yet how the mind combines sensory input into coherent crossmodal representations - the crossmodal binding problem - remains poorly understood. Here, we applied multi-echo fMRI across a four-day paradigm, in which participants learned 3-dimensional crossmodal representations created from well-characterized unimodal visual shape and sound features. Our novel paradigm decoupled the learned crossmodal object representations from their baseline unimodal shapes and sounds, thus allowing us to track the emergence of crossmodal object representations as they were learned by healthy adults. Critically, we found that two anterior temporal lobe structures - temporal pole and perirhinal cortex - differentiated learned from non-learned crossmodal objects, even when controlling for the unimodal features that composed those objects. These results provide evidence for integrated crossmodal object representations in the anterior temporal lobes that were different from the representations for the unimodal features. Furthermore, we found that perirhinal cortex representations were by default biased towards visual shape, but this initial visual bias was attenuated by crossmodal learning. Thus, crossmodal learning transformed perirhinal representations such that they were no longer predominantly grounded in the visual modality, which may be a mechanism by which object concepts gain their abstraction.