Predictive-coding and dynamical-systems accounts of mouse cortical activity that lean on recurrent E→E loops
This section is represented as a source-backed SciDEX wiki artifact. The full source remains in the original computational-review repository.
Source: https://github.com/AllenNeuralDynamics/ComputationalReviewRecurrence/blob/79ce062d54a924ce05953ec90aa9d26044d2b48f/content/14_predictive_coding.md
Citation anchors captured: 112
Citation contexts
-
1Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
7Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
8Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
9Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an... -
10CitationPredictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...content/14_predictive_coding.md:line 9Open reference Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...
-
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 Predictive coding’s most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 Latency of prediction-error / deviance signals in mouse V1. Onset latency (ms post-deviant or post-expected event) of reported prediction-error signals across three mouse V1 studies. Marker shape encodes study and population definition. The shaded band marks a ≈30–80 ms cortico-cortical loop window typical of V1↔higher-area interactions in mouse cortex. Paradigms differ (sequence-learning suppression vs. oddball... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 Latency of prediction-error / deviance signals in mouse V1. Onset latency (ms post-deviant or post-expected event) of reported prediction-error signals across three mouse V1 studies. Marker shape encodes study and population definition. The shaded band marks a ≈30–80 ms cortico-cortical loop window typical of V1↔higher-area interactions in mouse cortex. Paradigms differ (sequence-learning suppression vs. oddball... -
2Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 Latency of prediction-error / deviance signals in mouse V1. Onset latency (ms post-deviant or post-expected event) of reported prediction-error signals across three mouse V1 studies. Marker shape encodes study and population definition. The shaded band marks a ≈30–80 ms cortico-cortical loop window typical of V1↔higher-area interactions in mouse cortex. Paradigms differ (sequence-learning suppression vs. oddball... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 3Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 3Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 3Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 3Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
3Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 3Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 4Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 4Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. 4Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 4Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
4Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 4Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 5Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 5Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 5Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 5Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
5Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 5Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 6Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 6Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 6Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 6Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
6Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 6Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
7Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu... -
7Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 argue that unexpected visual stimuli boost the L2/3 V1 neurons most selective for that stimulus, supported by a pulvinar-VIP-SOM disinhibitory motif — a dedicated, feature-selective prediction-error signal. 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 reach the opposite conclusion: visuomotor “mismatch” signals in mouse V1 can be reproduced by purely sensory perturbations and are explained by feature selectivity plus convergence of... -
7Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 argue that unexpected visual stimuli boost the L2/3 V1 neurons most selective for that stimulus, supported by a pulvinar-VIP-SOM disinhibitory motif — a dedicated, feature-selective prediction-error signal. 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 reach the opposite conclusion: visuomotor “mismatch” signals in mouse V1 can be reproduced by purely sensory perturbations and are explained by feature selectivity plus convergence of... -
7Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 7Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 argue that unexpected visual stimuli boost the L2/3 V1 neurons most selective for that stimulus, supported by a pulvinar-VIP-SOM disinhibitory motif — a dedicated, feature-selective prediction-error signal. 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 reach the opposite conclusion: visuomotor “mismatch” signals in mouse V1 can be reproduced by purely sensory perturbations and are explained by feature selectivity plus convergence of... -
8Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
8Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
8Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 8Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference9 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
9Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference0 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference1 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference2 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
9Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference3 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference4 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference5 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
9Citation{ref}
sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference6 The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference7 identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while 9Citation{ref}sec-attractor-network-modelsevaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...content/14_predictive_coding.md:line 5Open reference8 showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off.... -
... 62 additional anchors in refs_json
References
- [Keller2018] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Shipp2024] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Shenoy2011] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Churchland2024] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Bennett2020] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Heilbron2026] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Keller2012] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Burkhalter2023] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Miyashita2024] “{ref}`sec-attractor-network-models` evaluated content-addressable and continuous attractor descriptions against the mouse record. Two complementary theoretical lenses now claim the same circuit motif — dense E→E recurrence within and between mouse cortical areas — but ask different questions of it. Predictive coding casts recurrence as a substrate for sending top-down predictions and reading out residual errors, an...”
- [Parras2017] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [LaoRodriguez2024] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Price2023] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Gallimore2023a] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Knudstrup2024] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Tang2023a] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Luo2023] “Predictive coding's most testable claim in mouse V1 is that learned regularities generate stimulus-selective responses to violations. The empirical record now contains four classes of evidence — late suppression of expected stimuli, deviance detection in oddball sequences, omission-evoked firing, and feature-selective amplification of unexpected stimuli — each measured in a different paradigm, and each examined in t...”
- [Knudstrup2025] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Becker2026] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Hamm2016] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Goltstein2018] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Peterka2026] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Wyrick2023] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Kirchberger2023] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Shin2023b] “The L2/3 deviance signals at 100–230 ms exceed the timing of a single feedforward sweep but fit comfortably within one to two cortico-cortical loop closures, an interpretation supported by causal feedback perturbations. [Knudstrup2025] showed that mouse V1 deviance responses depend on both first- and second-order sequence structure, scale with predictability, and exceed what short-term adaptation can explain; [Becke...”
- [Furutachi2024] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Fu2015] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Roth2016] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Huang2025b] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Fisek2023] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Larkum2004] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Kok2016] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Jordan2023] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Vasilevskaya2023] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Hoffmann2024] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Pakan2016] “Predictive coding in its Rao–Ballard form predicts that distinct cell classes carry predictions and errors, with feedforward and feedback laminar pathways distinguishing the two. The cell-class identity of error coders in mouse V1 remains the most actively contested element of this microcircuit story. [Furutachi2024] recorded mouse V1 L2/3 with pulvinar and interneuron perturbations and reported that unexpected visu...”
- [Muzzu2021] “[Furutachi2024] argue that unexpected visual stimuli boost the L2/3 V1 neurons most selective for that stimulus, supported by a pulvinar-VIP-SOM disinhibitory motif — a dedicated, feature-selective prediction-error signal. [Muzzu2021] reach the opposite conclusion: visuomotor "mismatch" signals in mouse V1 can be reproduced by purely sensory perturbations and are explained by feature selectivity plus convergence of...”
- [Jordan2020] “[Furutachi2024] argue that unexpected visual stimuli boost the L2/3 V1 neurons most selective for that stimulus, supported by a pulvinar-VIP-SOM disinhibitory motif — a dedicated, feature-selective prediction-error signal. [Muzzu2021] reach the opposite conclusion: visuomotor "mismatch" signals in mouse V1 can be reproduced by purely sensory perturbations and are explained by feature selectivity plus convergence of...”
- [Audette2023] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
- [Holey2024] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
- [Zhou2024] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
- [Raltschev2025] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
- [Shymkiv2025] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
- [Ding2026c] “The same predictive-coding logic has been tested in mouse auditory cortex with comparable partial success. [Audette2023] identified specific prediction-error signals in mouse auditory cortex and argued that errors may be computed early in sensory processing, while [Holey2024] showed that when the acoustic consequences of a movement become predictable, M2 responses to self-generated sounds are selectively gated off....”
Sister wikis (recently updated · no domain on this page)
- Agent Recipe: AI-for-Biology Closed-Loop with Reviewer Handoffs and Eval Contracts
- Agent Recipe: AI-for-Biology Closed-Loop with Reviewer Handoffs and Eval Contracts
- test
- JGBO-I27: Top 10 GBO Questions for Prioritization
- JGBO-I27: Top 10 GBO Questions for Prioritization
- Design Brief: Beta-test Evaluation Protocol for SciDEX v2 Design Trajectories
- Andy — Showcase Findings (auto-curated)
- Kris — Showcase Findings (auto-curated)
Recent activity here
No recent events touching this page.