Encoding of Stimulus Probability in Macaque Inferior Temporal Cortex
Open accessAbstract
Optimal perceptual decisions require sensory signals to be combined with prior information about stimulus probability. Although several theories propose that probabilistic information about stimulus occurrence is encoded in sensory cortex, evidence from neuronal recordings has not yet fully supported this view. We recorded activity from single neurons in inferior temporal cortex (IT) while monkeys performed a task that involved discriminating degraded images of faces and fruit. The relative probability of the cue being a face versus a fruit was manipulated by a latent variable that was not revealed to the monkeys and that changed unpredictably over the course of each recording session. In addition to responding to stimulus identity (face or fruit), population responses in IT encoded the long-term stimulus probability of whether a face or a fruit stimulus was more likely to occur. Face-responsive neurons showed reduced firing rates to expected faces, an effect consistent with "expectation suppression," but expected stimuli were decoded from multivariate population signals with greater accuracy. These findings support "predictive coding" theories, whereby neural signals in the mammalian visual system actively encode and update predictions about the local sensory environment.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Other studies have found that global suppression is attributable to suppression of neurons tuned away from the expected stimulus feature, while the representation of the expected stimulus is actually sharpened.
“others have found that this global suppression is attributable to the suppression of neurons tuned away from the expected stimulus feature, while the representation of the expected stimulus is actually sharpened”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
The Results (Neuronal responses to expected vs. unexpected stimuli; page unavailable) report reduced firing for expected faces, and the Discussion reports more accurate decoding of expected stimuli. But the Results (Multivariate analyses; page unavailable) link the strongest expectation-related response differences to neurons most selective for faces, not to suppression of neurons tuned away from the expected feature. The proposed explanation for global suppression is therefore not supported. (LLM, conf=0.86)
Experimental — accuracy not yet measuredOnly prediction signals, but not prediction errors, can precede sensory input.
“only prediction signals, but not prediction errors, can precede sensory input [87,128]”
From How Do Expectations Shape Perception? (2018) · cites this paper as [87]
A series of single- and multiunit recording studies have reported evidence of ES in macaque IT neurons.
“A series of single‐ and multiunit recording studies have also reported evidence of ES in macaque IT neurons. 85 , 88 , 89 , 90 , 91 , 92”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
Bell et al. found reduced population response to expected faces in face-selective IT neurons.
“The results showed a reduced population response to expected faces in face‐selective IT neurons, which Bell et al. interpreted as evidence that IT neurons encode prediction errors and long‐term probabilistic information, in line with PP.”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
ES effects in macaque IT neurons emerge at early latencies (~150 ms).
“These effects were sensitive to transitional statistics (e.g., the transition from Stimulus A to Stimulus B being more probable than from B to A), 90 , 91 persisted after controlling for RS, 88 , 89 and emerged at early latencies (∼150 ms). 88 , 90 , 91 , 92”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
ES effects in macaque IT neurons persist after controlling for RS.
“These effects were sensitive to transitional statistics (e.g., the transition from Stimulus A to Stimulus B being more probable than from B to A), 90 , 91 persisted after controlling for RS, 88 , 89 and emerged at early latencies (∼150 ms). 88 , 90 , 91 , 92”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
Because ES is typically calculated by comparing expected and unexpected stimulus responses, it is often unclear whether it represents genuine suppression, suppression relative to enhanced deviant responses, or both.
“as ES is typically calculated by comparing expected and unexpected stimulus responses, it is often unclear if it represents genuine suppression, suppression relative to the enhanced response to deviant stimuli, or both.”
From Evaluating the neurophysiological evidence for predictive processing as a model of perception (2020) · cites this paper as [88]
Expectations weakened response amplitude in macaque inferior temporal cortex but improved decoding performance.
“Similarly, a recent study found that expectations weakened response amplitude in macaque inferior temporal cortex (IT) but improved decoding performance [87].”
From How Do Expectations Shape Perception? (2018) · cites this paper as [87]
Model-based analyses that use computational models of perceptual inference or learning to derive trial-by-trial quantitative estimates of variables of interest and compare those to the data are one way to make more precise comparisons.
“One way to make more precise comparisons is to use model-based analyses that use computational models of perceptual inference or learning (e.g., [130]) to derive trial-by-trial quantitative estimates of variables of interest, and compare those to the data [87,128,131].”
From How Do Expectations Shape Perception? (2018) · cites this paper as [87]
Visuomotor integration, visual/auditory speech perception, and visual perception all benefit when sensory inputs are predictable.
“Visuomotor integration, visual/auditory speech perception, and visual perception all benefit when sensory inputs are predictable (1–3).”
From Layer and rhythm specificity for predictive routing (2020) · cites this paper as [2]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 48 · Cited by 118
- Authors
- Bell, A. H., Summerfield, C., Morin, E. L., Malecek, N. J., & Ungerleider, L. G.
- Affiliations
- National Institutes of Health (US)
- University of Oxford (GB)
- Corresponding author
- Bell, Andrew H.
- Venue
- Current Biology
- Publisher
- Elsevier BV
- Published
- 2016-09
- Volume
- 26
- Issue
- 17
- Pages
- 2280-2290
- ISSN
- 0960-9822 (ISSN-L)
- PMC
- PMC5021632
- OpenAlex
- W2513914902
- Cited by
- 118
- References
- 48
- Citations by year
- 2026: 6 · 2025: 17 · 2024: 16 · 2023: 15 · 2022: 11 · 2021: 9 · 2020: 12 · 2019: 8 · 2018: 11 · 2017: 12 · 2016: 1
- FWCI
- 6.79
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- Locations
- Current Biology (publishedVersion)
- PubMed (publishedVersion)
- Europe PMC (PubMed Central) (acceptedVersion)
- PubMed Central (acceptedVersion)
- Funders
- National Institute of Mental Health
- European Research Council (ERC) (281628)
- Topics
- Life Sciences › Neuroscience › Cognitive Neuroscience › Face Recognition and Perception
- Life Sciences › Neuroscience › Cognitive Neuroscience › Visual perception and processing mechanisms
- Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
- Keywords
- stimulus probability, inferior temporal cortex, predictive coding, prior information, face recognition (cognition), degraded images, population decoding, single-unit activity, macaque visual cortex, perceptual decision making
- MeSH terms
- Animals, Macaca mulatta, Male, Pattern Recognition, Visual, Temporal Lobe
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
