Where is the error? Hierarchical predictive coding through dendritic error computation
Open accessThe DOI matches its Crossref record; 58 of the 58 references it cites were found in a registry or an index.
Abstract
Top-down feedback in cortex is critical for guiding sensory processing, which has prominently been formalized in the theory of hierarchical predictive coding (hPC). However, experimental evidence for error units, which are central to the theory, is inconclusive and it remains unclear how hPC can be implemented with spiking neurons. To address this, we connect hPC to existing work on efficient coding in balanced networks with lateral inhibition and predictive computation at apical dendrites. Together, this work points to an efficient implementation of hPC with spiking neurons, where prediction errors are computed not in separate units, but locally in dendritic compartments. We then discuss the correspondence of this model to experimentally observed connectivity patterns, plasticity, and dynamics in cortex.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Emergence of simple-cell receptive field properties by learning a sparse code for natural images
Olshausen, B. A., & Field, D. J. (1996) · Nature · doi:10.1038/381607a0 · Cited by 5,694 · cited 1×
“…for both, learning and inference, where prediction errors are computed in tightly balanced dendritic compartments. neuroscience, as models for the sparse and efficient encoding of sensory data [13, 19, 20, 21] and as divisive normalization models of cortical computation [22, 23] (see also [24]).”
Rao, R. P. N., & Ballard, D. H. (1999) · Nature Neuroscience · doi:10.1038/4580 · Cited by 5,483 · cited 10×
“In the following we will focus on the less discussed effects of mismatch responses in visual cortex [78] and top-down modulated extra-classical receptive field effects [6], and outline how these types of responses can emerge in the prediction neurons of dendritic hPC.”
InfluentialInterneurons of the neocortical inhibitory system
Markram, H., Toledo-Rodriguez, M., Wang, Y., Gupta, A., Silberberg, G., & Wu, C. (2004) · Nature Reviews Neuroscience · doi:10.1038/nrn1519 · Cited by 2,954
Driving fast-spiking cells induces gamma rhythm and controls sensory responses
Cardin, J. A., Carlén, M., Meletis, K., Knoblich, U., Zhang, F., Deisseroth, K., Tsai, L.-H., & Moore, C. I. (2009) · Nature · doi:10.1038/nature08002 · Cited by 2,603
The Bayesian brain: The role of uncertainty in neural coding and computation
Knill, D. C., & Pouget, A. (2004) · Trends in Neurosciences · doi:10.1016/j.tins.2004.10.007 · Cited by 2,588
Canonical Microcircuits for Predictive Coding
Bastos, A. M., Usrey, W. M., Adams, R. A., Mangun, G. R., Fries, P., & Friston, K. J. (2012) · Neuron · doi:10.1016/j.neuron.2012.10.038 · Cited by 2,421 · cited 8×
“…support this idea: Layer 2/3 pyramidal neurons would send predictions, directly or mediated by layer 4 neurons, up the hierarchy where they are integrated at basal dendrites, and, directly or mediated by layer 5 neurons, down the hierarchy where they are integrated at apical dendrites [4, 17, 24].”
InfluentialGABAergic Interneurons in the Neocortex: From Cellular Properties to Circuits
Tremblay, R., Lee, S., & Rudy, B. (2016) · Neuron · doi:10.1016/j.neuron.2016.06.033 · Cited by 2,033 · cited 4×
“These are strong requirements, but indeed, cortex exhibits a wide array of inhibitory interneurons, some of which resemble the suggested connectivity patterns [72, 73].”
InfluentialNormalization as a canonical neural computation
Carandini, M., & Heeger, D. J. (2012) · Nature Reviews Neuroscience · doi:10.1038/nrn3136 · Cited by 1,906 · cited 1×
“…for both, learning and inference, where prediction errors are computed in tightly balanced dendritic compartments. neuroscience, as models for the sparse and efficient encoding of sensory data [13, 19, 20, 21] and as divisive normalization models of cortical computation [22, 23] (see also [24]).”
NEURONAL CIRCUITS OF THE NEOCORTEX
Douglas, R. J., & Martin, K. A. C. (2004) · Annual Review of Neuroscience · doi:10.1146/annurev.neuro.27.070203.144152 · Cited by 1,886
Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition
Gerstner, W., Kistler, W. M., Naud, R., & Paninski, L. (2014) · doi:10.1017/cbo9781107447615 · Cited by 1,515 · cited 1×
“This model is a special case of the spike response model with escape noise [41], and can be implemented by a standard leaky integrate and fire neuron with a noisy membrane potential.”
Hierarchical Bayesian inference in the visual cortex
Lee, T. S., & Mumford, D. (2003) · Journal of the Optical Society of America A · doi:10.1364/josaa.20.001434 · Cited by 1,473 · cited 1×
“…light of this success, it has been proposed that the primary computation that is performed by the cortex is a hierarchically organized inference process, where cortical areas combine bottom-up sensory information and top-down expectations to find a consistent explanation of sensory data [3, 4, 5].”
The neocortical circuit: themes and variations
Harris, K. D., & Shepherd, G. M. G. (2015) · Nature Neuroscience · doi:10.1038/nn.3917 · Cited by 1,215 · cited 4×
“More connections (especially within layer 2/3 between areas) might exist [17], but are left out here for simplicity.”
InfluentialCortical interneurons that specialize in disinhibitory control
Pi, H.-J., Hangya, B., Kvitsiani, D., Sanders, J. I., Huang, Z. J., & Kepecs, A. (2013) · Nature · doi:10.1038/nature12676 · Cited by 1,136
How Do Expectations Shape Perception?
de Lange, F. P., Heilbron, M., & Kok, P. (2018) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2018.06.002 · Cited by 975 · cited 3×
“A long line of research aims to describe these expectation-guided computations in our brain by Bayesian inference, i.e. statistically optimal perception [2], and subsequently could show that Bayesian inference can capture perception extraordinarily well [1].”
A Cortical Circuit for Gain Control by Behavioral State
Fu, Y., Tucciarone, J. M., Espinosa, J. S., Sheng, N., Darcy, D. P., Nicoll, R. A., Huang, Z. J., & Stryker, M. P. (2014) · Cell · doi:10.1016/j.cell.2014.01.050 · Cited by 937
Predictive Processing: A Canonical Cortical Computation
Keller, G. B., & Mrsic-Flogel, T. D. (2018) · Neuron · doi:10.1016/j.neuron.2018.10.003 · Cited by 836 · cited 2×
“To resolve this, it was proposed that errors are represented by deviations relative to a baseline firing rate [16], but this would require high firing rates and therefore seems implausible considering the low firing rates in neocortex [39].”
Long-range and local circuits for top-down modulation of visual cortex processing
Zhang, S., Xu, M., Kamigaki, T., Hoang Do, J. P., Chang, W.-C., Jenvay, S., Miyamichi, K., Luo, L., & Dan, Y. (2014) · Science · doi:10.1126/science.1254126 · Cited by 804
A disinhibitory circuit mediates motor integration in the somatosensory cortex
Lee, S., Kruglikov, I., Huang, Z. J., Fishell, G., & Rudy, B. (2013) · Nature Neuroscience · doi:10.1038/nn.3544 · Cited by 758
A cellular mechanism for cortical associations: an organizing principle for the cerebral cortex
Larkum, M. (2013) · Trends in Neurosciences · doi:10.1016/j.tins.2012.11.006 · Cited by 755 · cited 1×
“In a more general context it has been proposed that top-down connections could provide these predictions by targeting apical dendrites [14, 25, 26] (a similar architecture has also been used to implement backprop in a cortical microcircuit [27], see Infobox 2.2 for a comparison).”
InfluentialA neural circuit for spatial summation in visual cortex
Adesnik, H., Bruns, W., Taniguchi, H., Huang, Z. J., & Scanziani, M. (2012) · Nature · doi:10.1038/nature11526 · Cited by 702
Statistically optimal perception and learning: from behavior to neural representations
Fiser, J., Berkes, P., Orbán, G., & Lengyel, M. (2010) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2010.01.003 · Cited by 672 · cited 1×
“A long line of research aims to describe these expectation-guided computations in our brain by Bayesian inference, i.e. statistically optimal perception [2], and subsequently could show that Bayesian inference can capture perception extraordinarily well [1].”
Sensorimotor mismatch signals in primary visual cortex of the behaving mouse
Keller, G. B., Bonhoeffer, T., & Hübener, M. (2012) · Neuron · doi:10.1016/j.neuron.2012.03.040 · Cited by 668
Ferguson, B. R., & Gao, W.-J. (2018) · Frontiers in Neural Circuits · doi:10.3389/fncir.2018.00037 · Cited by 594
Neural substrates of vocalization feedback monitoring in primate auditory cortex
Eliades, S. J., & Wang, X. (2008) · Nature · doi:10.1038/nature06910 · Cited by 479
Efficient codes and balanced networks
Denève, S., & Machens, C. K. (2016) · Nature Neuroscience · doi:10.1038/nn.4243 · Cited by 471 · cited 9×
“That excitatory and inhibitory currents are balanced in cortical neurons is a well-established result [9], and it has been found in several instances that inhibition and excitation are tightly correlated, with inhibition trailing excitation by few milliseconds [9, 44, 45].”
InfluentialDoes predictive coding have a future?
Friston, K. (2018) · Nature Neuroscience · doi:10.1038/s41593-018-0200-7 · Cited by 451
Unique functional properties of somatostatin-expressing GABAergic neurons in mouse barrel cortex
Gentet, L. J., Kremer, Y., Taniguchi, H., Huang, Z. J., Staiger, J. F., & Petersen, C. C. H. (2012) · Nature Neuroscience · doi:10.1038/nn.3051 · Cited by 438
Great Expectations: Is there Evidence for Predictive Coding in Auditory Cortex?
Heilbron, M., & Chait, M. (2018) · Neuroscience · doi:10.1016/j.neuroscience.2017.07.061 · Cited by 428
Theories of Error Back-Propagation in the Brain
Whittington, J. C. R., & Bogacz, R. (2019) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2018.12.005 · Cited by 420 · cited 1×
“To this end an ’inverted’ model of hPC is employed [29], where inhibition at the apical dendrite is used to compute the backpropagated error of the output. balance can be combined in a single model with dendritic error computation [13, 35, 36], relating these concepts to the proposed model of…”
Fine-scale specificity of cortical networks depends on inhibitory cell type and connectivity
Yoshimura, Y., & Callaway, E. M. (2005) · Nature Neuroscience · doi:10.1038/nn1565 · Cited by 390
Buesing, L., Bill, J., Nessler, B., & Maass, W. (2011) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1002211 · Cited by 384 · cited 2×
“…is to point out the intimate relations that exist between three previously mostly independent branches of research of theoretical neuroscience: Hierarchical predictive coding [6], efficient coding in balanced spiking networks [9] and the theory of neural sampling in probabilistic models [10].”
Suboptimality in perceptual decision making
Rahnev, D., & Denison, R. N. (2018) · Behavioral and Brain Sciences · doi:10.1017/s0140525x18000936 · Cited by 342
Clustered dynamics of inhibitory synapses and dendritic spines in the adult neocortex.
Chen, J. L., Villa, K. L., Cha, J. W., So, P. T. C., Kubota, Y., & Nedivi, E. (2012) · Neuron · doi:10.1016/j.neuron.2012.02.030 · Cited by 341 · cited 1×
“dendrites, where it is known that correlated excitatory synapses cluster together [96, 97, 98].”
Liu, G. (2004) · Nature Neuroscience · doi:10.1038/nn1206 · Cited by 339 · cited 3×
“Crucially, this local balance can be observed down to the scale of (at least) single dendritic branches [47], since the attenuation of dendritic currents prevents that inhibitory postsynaptic potentials spread into other dendritic compartments and influence the E/I balance there [50, 51].”
Evaluating the neurophysiological evidence for predictive processing as a model of perception
Walsh, K. S., McGovern, D. P., Clark, A., & O'Connell, R. G. (2020) · Annals of the New York Academy of Sciences · doi:10.1111/nyas.14321 · Cited by 333 · cited 5×
“How some of these effects can occur in models without error units, such as omission responses or expectation suppression, has been discussed before [5, 43].”
InfluentialExperience-dependent spatial expectations in mouse visual cortex
Fiser, A., Mahringer, D., Oyibo, H. K., Petersen, A. V., Leinweber, M., & Keller, G. B. (2016) · Nature Neuroscience · doi:10.1038/nn.4385 · Cited by 332 · cited 1×
“This interpretation of mismatch responses could explain why mismatch responses in V1 seem to be mixed with prediction responses [78, 84].”
Cellular Mechanisms of Conscious Processing
Aru, J., Suzuki, M., & Larkum, M. E. (2020) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2020.07.006 · Cited by 321
Bourne, J. N., & Harris, K. M. (2011) · Hippocampus · doi:10.1002/hipo.20768 · Cited by 318
Free Energy, Precision and Learning: The Role of Cholinergic Neuromodulation
Moran, R. J., Campo, P., Symmonds, M., Stephan, K. E., Dolan, R. J., & Friston, K. J. (2013) · The Journal of Neuroscience · doi:10.1523/jneurosci.4255-12.2013 · Cited by 312
With or without you: predictive coding and Bayesian inference in the brain
Aitchison, L., & Lengyel, M. (2017) · Current Opinion in Neurobiology · doi:10.1016/j.conb.2017.08.010 · Cited by 308 · cited 5×
“How some of these effects can occur in models without error units, such as omission responses or expectation suppression, has been discussed before [5, 43].”
InfluentialGeneration of end-inhibition in the visual cortex via interlaminar connections
Bolz, J., & Gilbert, C. D. (1986) · Nature · doi:10.1038/320362a0 · Cited by 287
A tutorial on the free-energy framework for modelling perception and learning
Bogacz, R. (2017) · Journal of Mathematical Psychology · doi:10.1016/j.jmp.2015.11.003 · Cited by 283 · cited 2×
“…hPC was to represent prediction errors within a distinct neural population of error units, that integrate inputs of prediction units within the same level, and subtract top-down predictions according to where decoding weights D i now correspond directly to weights of neural connections [11].”
Cell-type homologies and the origins of the neocortex
Dugas-Ford, J., Rowell, J. J., & Ragsdale, C. W. (2012) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1204773109 · Cited by 278
Learning by the Dendritic Prediction of Somatic Spiking
Urbanczik, R., & Senn, W. (2014) · Neuron · doi:10.1016/j.neuron.2013.11.030 · Cited by 230 · cited 6×
“Therefore, these dendritic potentials can easily represent the prediction errors that are required for the learning of bottom-up and top-down connections (Infobox 2.1), which has been successfully applied in [13, 14].”
InfluentialHeeger, D. J. (2017) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1619788114 · Cited by 220 · cited 1×
“An open question is how these mechanisms can purposefully interact with the learning of predictions in a hierarchical model (but see [107] for a suggestion without learning).”
Functional connectomics spanning multiple areas of mouse visual cortex
The MICrONS Consortium, Bae, J. A., Baptiste, M., Bishop, C. A., Bodor, A. L., Brittain, D., Buchanan, J., Bumbarger, D. J., Castro, M. A., Celii, B., et al. (2021) · doi:10.1101/2021.07.28.454025 · Cited by 214
Kruglikov, I., & Rudy, B. (2008) · Neuron · doi:10.1016/j.neuron.2008.04.024 · Cited by 209
Predictive coding as a model of response properties in cortical area V1
Spratling, M. W. (2010) · The Journal of Neuroscience · doi:10.1523/jneurosci.4911-09.2010 · Cited by 206
GABA promotes the competitive selection of dendritic spines by controlling local Ca2+ signaling
Hayama, T., Noguchi, J., Watanabe, S., Takahashi, N., Hayashi-Takagi, A., Ellis-Davies, G. C. R., Matsuzaki, M., & Kasai, H. (2013) · Nature Neuroscience · doi:10.1038/nn.3496 · Cited by 193 · cited 3×
“One prediction of these derived VDP rules is that strong local inhibition should promote the depression of excitatory synapses, an effect that has been observed in proximal dendrites of hippocampal pyramidal neurons [64] (Fig 2; see also [67]).”
Mismatch Receptive Fields in Mouse Visual Cortex
Zmarz, P., & Keller, G. B. (2016) · Neuron · doi:10.1016/j.neuron.2016.09.057 · Cited by 178 · cited 3×
“In the following we will focus on the less discussed effects of mismatch responses in visual cortex [78] and top-down modulated extra-classical receptive field effects [6], and outline how these types of responses can emerge in the prediction neurons of dendritic hPC.”
Untangling GABAergic wiring in the cortical microcircuit
Kubota, Y. (2014) · Current Opinion in Neurobiology · doi:10.1016/j.conb.2013.10.003 · Cited by 172 · cited 2×
“These are strong requirements, but indeed, cortex exhibits a wide array of inhibitory interneurons, some of which resemble the suggested connectivity patterns [72, 73].”
Neocortical Layer 1: an elegant solution to top-down and bottom-up integration
Schuman, B., Dellal, S., Prönneke, A., Machold, R., & Rudy, B. (2021) · Annual Review of Neuroscience · doi:10.1146/annurev-neuro-100520-012117 · Cited by 148
Multistability and perceptual inference
Gershman, S. J., Vul, E., & Tenenbaum, J. B. (2012) · Neural Computation · doi:10.1162/neco_a_00226 · Cited by 140 · cited 1×
“Neural sampling therefore is a principled way to represent uncertainty in inference via neural activity, and has for example been used to explain variability in neural responses [37], and the origin of multistability in perception [38].”
Neocortical Evolution: Neuronal Circuits Arise Independently of Lamination
Karten, H. J. (2013) · Current Biology · doi:10.1016/j.cub.2012.11.013 · Cited by 126 · cited 1×
“This idea is supported by evidence from studies of brain anatomy and cell-type markers [109, 110, 111] (Fig 4A,B).”
Predictive Coding: a Theoretical and Experimental Review
Millidge, B., Seth, A., & Buckley, C. L. (2021) · arXiv · doi:10.48550/arxiv.2107.12979 · Cited by 106 · cited 3×
“Due to these difficulties, to the best of our knowledge, as of yet no complete implementation of hPC has been proposed that uses spiking error units (see also [8] for a recent and extensive review).”
Is predictive coding theory articulated enough to be testable?
Kogo, N., & Trengove, C. (2015) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2015.00111 · Cited by 98 · cited 1×
“However, classical hPC for the most part remains on the level of firing-rate dynamics of neural populations, and it has proven difficult to connect the theory to the properties of single neurons with spiking dynamics [7, 8].”
Precisely Timed Nicotinic Activation Drives SST Inhibition in Neocortical Circuits
Urban-Ciecko, J., Jouhanneau, J.-S., Myal, S. E., Poulet, J. F. A., & Barth, A. L. (2018) · Neuron · doi:10.1016/j.neuron.2018.01.037 · Cited by 93
Aitchison, L., & Lengyel, M. (2016) · PLOS Computational Biology · doi:10.1371/journal.pcbi.1005186 · Cited by 67 · cited 3×
“Second, noise in spiking neural networks enables neural sampling [10, 35, 36].”
Influential
Cited by
How deep is the brain? The shallow brain hypothesis
Suzuki, M., Pennartz, C. M. A., & Aru, J. (2023) · Nature Reviews Neuroscience · doi:10.1038/s41583-023-00756-z · Cited by 79
“174–179 for models that incorporate dendritic computational capacities).”
BackgroundPredictive coding networks for temporal prediction
Millidge, B., Tang, M., Osanlouy, M., Harper, N. S., & Bogacz, R. (2024) · PLOS Computational Biology · doi:10.1371/journal.pcbi.1011183 · Cited by 44
“Since this work focuses on linear tPC and its properties, we leave the exploration of these directions for future investigations.”
BackgroundPredictive coding with spiking neurons and feedforward gist signaling
Lee, K., Dora, S., Mejias, J. F., Bohte, S. M., & Pennartz, C. M. A. (2024) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2024.1338280 · Cited by 14
Learning beyond sensations: How dreams organize neuronal representations
Deperrois, N., Petrovici, M. A., Senn, W., & Jordan, J. (2024) · Neuroscience & Biobehavioral Reviews · doi:10.1016/j.neubiorev.2023.105508 · Cited by 9
Brucklacher, M., Pezzulo, G., Mannella, F., Galati, G., & Pennartz, C. M. A. (2025) · Neural Networks · doi:10.1016/j.neunet.2024.106716 · Cited by 5
Metadata
- Authors
- Mikulasch, F. A., Rudelt, L., Wibral, M., & Priesemann, V.
- Venue
- Trends in Neurosciences
- Publisher
- Elsevier BV
- Published
- 2023-01
- Volume
- 46
- Issue
- 1
- Pages
- 45-59
- ISSN
- 0166-2236 (ISSN-L)
- OpenAlex
- W4309475065
- Cited by
- 121
- References
- 58
- FWCI
- 3.34
- Type
- journal article
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.