Task learning increases information redundancy of neural responses in macaque visual cortex

Shizhao Liu; Anton Pletenev; Ralf M. Haefner; Adam C. Snyder

q-bio.NCMar 7, 2026

Task learning increases information redundancy of neural responses in macaque visual cortex

Authors: Shizhao Liu, Anton Pletenev, Ralf M. Haefner, Adam C. Snyder

Organizations: Department of Brain and Cognitive Sciences, University of Rochester, Rochester & 14627, USA. · Center for Visual Science, University of Rochester, Rochester & 14627, USA. · Department of Computer Science, University of Rochester, Rochester & 14627, USA · Department of Physics & Astronomy, University of Rochester, Rochester & 14627, USA ·

Abstract

How does the brain optimize sensory information for decision-making in new tasks? One hypothesis suggests learning reduces redundancy in neural representations to improve efficiency, while another, based on Bayesian inference, predicts learning increases redundancy by distributing information across neurons. We tested these hypotheses by tracking population responses in macaque cortical area V4 as monkeys learned visual discrimination tasks. We found strong support for the Bayesian predictions: task learning increased redundancy in neural responses over weeks of training and within single trials. This redundancy did not reduce information but instead increased the information carried by individual neurons. These insights suggest sensory processing in the brain reflects a generative rather than discriminative inference process.

Figures & tables

Fig. 1 : Illustration of experiment and key predictions. a : Response of two example neurons to a horizontal and a vertical stimulus of intermediate signal strength (“coherence”). Each neuron’s response is variable (middle), and this variability is correlated. In this example, positive correlations ( $\rho=0.32$ , Pearson correlation) align with the task-tuning axis (both neurons prefer 0 degree stimulus), reducing the decoding performance (right). b : Linear Fisher information as a function of population size for two example sessions: during the task early in learning, during the task late in learning, and during passive viewing late in learning. $I_{\rm real}$ (orange) denotes actual (linear) Fisher information, and $I_{\rm shuffle}$ (yellow) represents the linear Fisher information after shuffling trials, thereby destroying noise correlations. c : Differential predictions for changes in redundancy over the course of task learning.

Fig. 2 : Experimental design and behavior. a : Task design: after animals fixated the center to start a trial. Choice targets were shown, followed by the stimulus. After stimulus offset, animals indicated their choice by a saccade to one target. Bottom: study timeline. b : Example stimulus images for four orientations and two coherence levels. c : Spatial layout of the fixation point (blue dot), stimulus (dashed circle), and choice targets (black and white dots). The red numbers (not shown during the experiment) indicate the stimulus orientation corresponding to each choice target. d : Psychometric curves of late-learning sessions. e : Orientation kernels of late-learning sessions. Dashed lines: ideal observer predictions. Solid lines: across-sessions average. Shaded areas: the standard error of the mean across sessions. The bars on the right indicate the bias averaged across sessions, and the error bars indicate the standard error. (The temporal kernels in Fig. S26 ) f : Time course of the learning index for two animals and two tasks. Solid lines and shading represent medians and 68% confidence intervals across 1000 bootstrap samples. Gray vertical lines separate each epoch into “early-learning” and “late-learning”. (Time course with real experiment dates can be found in Fig. S27 )

Fig. 3 : Redundancy increases over the course of learning. a : Time course of

I_{\rm redundancy}

during the course of learning using sub-sampled populations with consistent number of units (monkey R: cardinal:

n=17

, oblique:

n=18

; monkey G: oblique

n=44

, cardinal

n=55

). Lines and error bars are means and 68% confidence interval across bootstrapped samples. b : Scatter plots between the learning index and

I_{\rm redundancy}

for two tasks and two animals. Each dot represents a session. Black dashed lines are linear regression fit across sessions. c : We separated sessions into two groups: “early” and “late”, using the boundary show in Fig. 2 f. The bar plots averaged

I_{\rm redundancy}

across sessions in each learning stage when the animals were performing the task or passively viewing the stimulus. Error bars indicate the standard error of the mean across sessions. (Significance levels:

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

)

Fig. 4 : Increase in redundancy is explained by increase in

I_{\rm shuffle}

, not a decrease in

I_{\rm real}

. a : Information sharing during generative inference. The activities of sensory neurons,

r_{i}

, represent posterior beliefs about the intensity of oriented edges in the input stimulus. A decision-making area computes a belief about a decision variable,

p(d)

. This belief, based on the information in all sensory neurons, informs all sensory neurons via feedback signals (blue arrows). b, c : Model [ 23 ] simulations at different stages of learning. Errorbars indicate 68% confidence interval across random samples of 32 (panel b ) and 8, 32, 256 (panel c ) out of a total of 256 model neurons. In panel c , for each population size, the values were normalized by the mean

I_{\rm real}

before learning. d : Time course of

I_{\rm real}

I_{\rm shuffle}

and behavioral Fisher information

I_{\textbf{behav}}

during learning. Shaded areas: 68% across Bootstrap samples. e : Correlation between learning index

I_{\rm real}

and

I_{\rm shuffle}

, respectively. One dot per session. Black lines are a linear regression fit across sessions, solid for

I_{\rm real}

and dashed for

I_{\rm shuffle}

. (Significance levels:

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

)

Fig. 5 : Redundancy increases within a trial. a : Temporal propagation of beliefs during generative inference.

t_{1}..t_{4}

represent times within a single trial. b : Change in

I_{\rm redundancy}

across eight time bins within a trial for three stages of learning (model). Error bars are 68% confidence interval across 256 random sub-samples (size = 32) of model neurons. c :

I_{\rm redundancy}

for each 200ms time bin within a trial in empirical data, controlled for population size. Solid line: average across late-learning sessions. Dashed line: average across early-learning sessions.

\beta

is the regression coefficient:

I_{\rm redundancy}

vs time bin. Error bars indicate the standard error of the mean. across sessions (Significance levels:

{}^{*}p<0.05

{}^{**}p<0.01

{}^{***}p<0.001

)

Fig. S1 : Correlation between learning index and behavioral accuracy across coherence levels. For each epoch, coherence levels that appeared in multiple sessions were selected. We examined the correlation between the behavioral accuracy of each coherence level and the learning index for each session. Overall, strong positive correlations were observed between these two measures of behavioral performance. Note that the learning index was calculated using data from multiple coherence levels, so deviations from

r=1

are expected. (*:

p<0.05

; **:

p<0.01

; ***:

p<0.001

Fig. S4 : Histograms of basic properties of analyzed units. In each panel, units from late-learning sessions are presented on top while units from early-learning sessions are at the bottom. The triangles indicate the median values. a : Averaged firing rates in response to all coherence levels. b : Tuning index computed with the highest coherence level for each animal (monkey R: 10%; monkey G: 15%). c : Stimulus $d^{\prime}$ computed with the highest coherence level for each animal. d : Choice $d^{\prime}$ , computed for each coherence level, then averaged across coherence levels. In panels b and c , the filled histograms mark individual units with significant stimulus tuning or choice tuning (independent samples t-test).

Fig. S5 : See caption on the next page

Fig. S7 :

I_{\rm redundancy}

computed using the whole neural populations over the course of learning. Same as Fig. 3 , but for the whole populations, and that the error bars in panel a indicate analytically computed standard deviations ( materials and methods ). Spearman rank correlations between

I_{\rm redundancy}

and the learning index are reported for each case: monkey R, cardinal task:

\rho(22)=0.71

p=

1.55\text{\times}{10}^{-04}

; monkey R, oblique task: $\rho(12)=0.72$ , $p=$4.81\text{\times}{10}^{-03}

; monkey G, oblique task:

\rho(46)=0.62

p=

4.62\text{\times}{10}^{-06}

; monkey G, cardinal task: $\rho(35)=0.67$ , $p=$1.18\text{\times}{10}^{-05}

. (*:

p<0.05

; **:

p<0.01

; ***:

p<0.001

)

Fig. S8 :

I_{\rm real}

and

I_{\rm shuffle}

as a function of neural population size in example sessions. One session before learning and one session after learning were selected for each task epoch. For each population size, units were bootstrapped 1,000 times or as many times as allowed by the total population size. Solid lines indicate the mean across samples, and shaded areas represent the 68% confidence intervals.

Fig. S9 : See caption on the next page

Fig. S11 : Information redundancy within population of high or low stimulus

d^{\prime}

based on task-performing data. Units from each session were separated into two halves based on their stimulus

d^{\prime}

( Methods ), and

I_{\rm redundancy}

was computed separately for each of the two populations, consisting of higher stimulus

d^{\prime}

units, and lower stimulus

d^{\prime}

units, respectively. Size of population was kept constant within each epoch (monkey R, cardinal: N = 8, monkey R, oblique: N = 9; monkey G, oblique: N = 22; monkey G cardinal: N = 27) a : Scatter plots between

I_{\rm redundancy}

of higher

d^{\prime}

/lower

d^{\prime}

population (represented by filled/empty symbols) and learning index. b : Scatter plots comparing

I_{\rm redundancy}

for higher stimulus

d^{\prime}

population with

I_{\rm redundancy}

for lower

d^{\prime}

population. Each dot represents a session. The late-learning sessions are shown in the top row while the early-learning sessions are in the bottom row.

Fig. S12 :

I_{\rm redundancy}

in passive-viewing data. Same as Fig. 3 but computed with neural responses during passive-viewing. Spearman rank correlation between

I_{\rm redundancy}

and the learning index: monkey R, cardinal:

\rho(11)=0.01

p=

0.978

; monkey R, oblique: $\rho(12)=0.74$ , $p=$3.82\text{\times}{10}^{-03}

; monkey G, oblique:

\rho(46)=0.06

p=

0.671

; monkey G, cardinal: $\rho(36)=0.02$ , $p=$0.919

. This figure shows that

I_{\rm redundancy}

in passive-viewing data did not consistently increase with learning.

Fig. S13 : Correlation between

I_{\rm redundancy}

and behavioral accuracy for individual coherence levels. For each epoch, a subset of coherence levels that appeared in multiple sessions was selected to examine the correlation between

I_{\rm redundancy}

and behavioral accuracy. Overall, positive correlations were observed, though some did not reach statistical significance, likely due to the limited number of sessions. This analysis supports the conclusion that the increase in

I_{\rm redundancy}

with learning, as shown in Fig. 3 , cannot be attributed to changes in coherence composition during learning. (*:

p<0.05

; **:

p<0.01

; ***:

p<0.001

)

Fig. S14 : Time course of eye dynamic variables. We examined five eye dynamic variables: position projected on the x-axis and y-axis, velocity along x-axis and y-axis, as well as pupil size. For each variable, we computed the mean values of each trial, then the average values (shown in the left two columns) and variance (the right two columns) across trials within each session. Error bars indicate 68% confidence interval estimated from 1000 Bootstraps of trials for each session. Position and velocity are in visual degrees, while pupil size is in arbitrary units and was normalized for each animal.

Fig. S15 : Decoding accuracy of animals’ choice from eye dynamic variables

Fig. S16 : See caption on next page.

Fig. S18 :

I_{\rm real}

and

I_{\rm shuffle}

in passive-viewing data. Same as Fig. 4 but computed with neural responses during passive-viewing. Spearman correlations between

I_{\rm real}

and learning index: monkey R cardinal:

\rho(11)=-0.36

p=

0.232

; monkey R, oblique: $\rho(12)=0.49$ , $p=$0.0809

; monkey G, oblique:

\rho(46)=-0.58

p=

2.53\text{\times}{10}^{-05}

; monkey G, cardinal: $\rho(36)=-0.03$ , $p=$0.876

. Spearman correlations between

I_{\rm shuffle}

and learning index: monkey R, cardinal:

\rho(11)=-0.47

p=

0.110

; monkey R, oblique: $\rho(12)=0.79$ , $p=$1.28\text{\times}{10}^{-03}

; monkey G, oblique:

\rho(46)=-0.34

p=

0.0176

; monkey G, cardinal: $\rho(36)=-0.09$ , $p=$0.591

. This figure shows that

I_{\rm real}

and

I_{\rm shuffle}

in passive-viewing data did not consistently increase with learning.

Fig. S19 : Percentage of redundant information

(I_{\rm redundancy}/I_{\rm shuffle}\times 100)

. a : Percentage of redundant information increased with training time. b : Scatter plot between this percentage and learning index. Spearman rank correlation: Monkey R, cardinal task:

\rho(21)=0.68

p=

4.90\text{\times}{10}^{-04}

; Monkey R, oblique task: $\rho(12)=0.52$ , $p=$6.16\text{\times}{10}^{-02}

; Monkey G, oblique task:

\rho(46)=0.58

p=

2.02\text{\times}{10}^{-05}

; Monkey G, cardinal task: $\rho(36)=0.52$ , $p=$1.09\text{\times}{10}^{-03}

Fig. S20 :

I_{\rm real}

and

I_{\rm shuffle}

over the course of learning computed using the whole neural populations. Same as Fig. 4 , but for the whole populations, and that the error bars in panel a indicate analytically computed standard deviations ( materials and methods ). Spearman correlation coefficients between

I_{\rm real}

and

I_{\rm behav}

are reported for each condition: monkey R, Cardinal:

\rho(22)=0.59

p=

2.62\text{\times}{10}^{-03}

; monkey R, Oblique: $\rho(12)=0.92$ , $p<$2.2\text{\times}{10}^{-308}

; monkey G, Cardinal:

\rho(35)=0.38

p=

0.0196

; monkey G, Oblique: $\rho(46)=0.06$ , $p=$0.688

. Spearman correlations between

I_{\rm shuffle}

and learning index: monkey R, cardinal:

\rho(22)=0.71

p=

1.43\text{\times}{10}^{-04}

; monkey R, oblique: $\rho(12)=0.76$ , $p=$0.002\,52

; monkey G, oblique:

\rho(46)=0.48

p=

7.35\text{\times}{10}^{-04}

; monkey G, cardinal: $\rho(35)=0.71$ , $p=$1.97\text{\times}{10}^{-06}

. Spearman correlations between

I_{\rm real}

and the learning index: monkey R, cardinal:

\rho(22)=0.64

p=

9.90\text{\times}{10}^{-04}

; monkey R, oblique: $\rho(12)=0.81$ , $p=$8.10\text{\times}{10}^{-04}

; monkey G, oblique:

\rho(46)=0.01

p=0.93

; monkey G, cardinal:

\rho(35)=0.32

p=

0.0553$$ .

Fig. S21 : Comparison of neurometric and psychometric thresholds. a : Time course of neurometric and psychometric thresholds throughout learning. b : Scatter plot of neurometric versus psychometric thresholds for well-performed sessions.

Fig. S24 :

I_{\rm redundancy}

for each 50 ms time bin within a trial in empirical data. Same as Fig. 5 except with finer temporal resolution of 50 ms and using the whole neural populations. Top: monkey R. Bottom: monkey G. Linear regression statistics: monkey R, cardinal task: Late-learning:

\beta=

-7.3466\text{\times}{10}^{-06}

, $p=$0.908\,62

; Early-learning:

\beta=

-5.9143\text{\times}{10}^{-05}

, $p=$0.344\,19

; monkey R, oblique task: Late-learning:

\beta=

9.4392\text{\times}{10}^{-05}

, $p=$0.002\,256\,8

; Early-learning:

\beta=

1.1676\text{\times}{10}^{-05}

, $p=$0.027\,074

; monkey G, oblique task: Late-learning:

\beta=

1.5672\text{\times}{10}^{-04}

, $p=$4.8610\text{\times}{10}^{-13}

; Early-learning:

\beta=

9.2195\text{\times}{10}^{-06}

, $p=$0.347\,47

; monkey G, cardinal task: Late-learning:

\beta=

2.3171\text{\times}{10}^{-04}

, $p=$6.5475\text{\times}{10}^{-39}

; Early-learning:

\beta=

-2.1638\text{\times}{10}^{-05}

, $p=$0.182\,88

Fig. S25 : Information redundancy within population of high or low stimulus

d^{\prime}

based on passive-viewing data. Same analysis as in figure S11 but now the units are split by their task sensitivity computed from passive viewing data, as opposed to task data as in figure S11 . Size of population was again kept constant within each epoch (monkey R, cardinal: N = 8, monkey R, oblique: N = 9; monkey G, oblique: N = 22; monkey G cardinal: N = 27).

Fig. S26 : Temporal kernels of each epoch. The lines represent the average across late-learning sessions, while the error bars indicate the standard error of the mean across these sessions.

Fig. S27 : Time course of behavioral performance with real calendar dates. The x-axis represents calendar days, with the spacing reflecting the true time interval (in days) between sessions. For the monkey G oblique epoch, two months of training sessions were excluded from the analysis due to the following reason: the animal was initially trained with the stimulus presented in the left hemifield but struggled to generalize when the stimulus was moved to the right hemifield after implantation surgery to his left V4. To aid in performance recovery, we modified the stimulus size and location during those sessions, leading to inconsistencies in visual stimulus parameters.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
$d^{\prime}$ during task
Correlation coefficients	$p=0.032$	$p=0.393$	$p=0.014$	$p=0.098$
Regression coefficients	$p=0.023$	$p=0.007$	$p<0.0001$	$p=0.021$
$d^{\prime}$ during passive viewing
Correlation coefficients	$p=0.451$	$p=0.260$	$p=0.937$	$p=0.290$
Regression coefficients	$p=0.521$	$p=0.007$	$p=0.660$	$p=0.461$

Table S1 :

p-

values from permutation-based test that compares the relationship between

I_{\rm redundancy}

and learning index for higher stimulus

d^{\prime}

units and that for lower

d^{\prime}

units.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
$d^{\prime}$ during task
Late-learning	$t(16)=3.98$ $p=$ 1.07\text{\times}{10}^{-03}$$	$t(8)=5.90$ $p=$ 3.61\text{\times}{10}^{-04}$$	$t(22)=3.35$ $p=$ 2.91\text{\times}{10}^{-03}$$	$t(26)=3.05$ $p=$ 5.22\text{\times}{10}^{-03}$$
Early-learning	$t(5)=0.40$ $p=$ 7.02\text{\times}{10}^{-01}$$	$t(4)=1.73$ $p=$ 1.58\text{\times}{10}^{-01}$$	$t(24)=-4.38$ $p=$ 2.00\text{\times}{10}^{-04}$$	$t(10)=1.10$ $p=$ 2.97\text{\times}{10}^{-01}$$
$d^{\prime}$ during passive viewing
Late-learning	$t(12)=4.90$ $p=$ 3.68\text{\times}{10}^{-04}$$	$t(8)=5.45$ $p=$ 6.05\text{\times}{10}^{-04}$$	$t(22)=-2.40$ $p=$ 2.52\text{\times}{10}^{-02}$$	$t(26)=-0.70$ $p=$ 4.92\text{\times}{10}^{-01}$$
Early-learning	No $d^{\prime}$ data	$t(4)=1.52$ $p=$ 2.03\text{\times}{10}^{-01}$$	$t(24)=-3.75$ $p=$ 9.92\text{\times}{10}^{-04}$$	$t(10)=-2.40$ $p=$ 3.71\text{\times}{10}^{-02}$$

Table S2 : Paired t-test that compares

I_{\rm redundancy}

within higher

d^{\prime}

units with that within lower

d^{\prime}

units

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
x-pos-avg	$r(20)=0.35$ , $p=$ 1.15\text{\times}{10}^{-01}$$	$r(12)=0.09$ , $p=$ 7.50\text{\times}{10}^{-01}$$	$r(46)=-0.23$ , $p=$ 1.12\text{\times}{10}^{-01}$$	$r(35)=-0.01$ , $p=$ 9.40\text{\times}{10}^{-01}$$
y-pos-avg	$r(20)=-0.37$ , $p=$ 9.41\text{\times}{10}^{-02}$$	$r(12)=0.11$ , $p=$ 7.04\text{\times}{10}^{-01}$$	$r(46)=-0.21$ , $p=$ 1.55\text{\times}{10}^{-01}$$	$r(35)=-0.00$ , $p=$ 9.77\text{\times}{10}^{-01}$$
x-vel-avg	$r(20)=-0.25$ , $p=$ 2.60\text{\times}{10}^{-01}$$	$r(12)=-0.20$ , $p=$ 4.83\text{\times}{10}^{-01}$$	$r(46)=-0.12$ , $p=$ 4.28\text{\times}{10}^{-01}$$	$\mathbf{r(35)=0.44}$ , $p=$ 7.42\text{\times}{10}^{-03}$$
y-vel-avg	$r(20)=0.16$ , $p=$ 4.63\text{\times}{10}^{-01}$$	$r(12)=-0.26$ , $p=$ 3.74\text{\times}{10}^{-01}$$	$\mathbf{r(46)=0.65}$ , $p=$ 1.04\text{\times}{10}^{-06}$$	$r(35)=0.03$ , $p=$ 8.74\text{\times}{10}^{-01}$$
pupil-avg	$r(20)=0.23$ , $p=$ 3.05\text{\times}{10}^{-01}$$	$r(12)=0.23$ , $p=$ 4.36\text{\times}{10}^{-01}$$	$\mathbf{r(46)=-0.48}$ , $p=$ 6.05\text{\times}{10}^{-04}$$	$\mathbf{r(35)=-0.62}$ , $p=$ 6.26\text{\times}{10}^{-05}$$
x-pos-var	$\mathbf{r(20)=0.80}$ , $p=$ 7.54\text{\times}{10}^{-06}$$	$r(12)=-0.50$ , $p=$ 6.94\text{\times}{10}^{-02}$$	$\mathbf{r(46)=0.70}$ , $p=$ 1.19\text{\times}{10}^{-07}$$	$\mathbf{r(35)=0.47}$ , $p=$ 3.81\text{\times}{10}^{-03}$$

Table S3 : Spearman correlation between learning index and eye dynamic variables. Significant correlations (

p<0.05

) are marked bold.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
x-pos-avg	$r(19)=0.70$ , $p=$ 4.06\text{\times}{10}^{-04}$$	$r(11)=0.76$ , $p=$ 2.78\text{\times}{10}^{-03}$$	$r(45)=0.61$ , $p=$ 5.53\text{\times}{10}^{-06}$$	$r(34)=0.66$ , $p=$ 1.15\text{\times}{10}^{-05}$$
y-pos-avg	$r(19)=0.69$ , $p=$ 4.77\text{\times}{10}^{-04}$$	$r(11)=0.70$ , $p=$ 7.67\text{\times}{10}^{-03}$$	$r(45)=0.59$ , $p=$ 1.46\text{\times}{10}^{-05}$$	$r(34)=0.67$ , $p=$ 8.84\text{\times}{10}^{-06}$$
x-vel-avg	$r(19)=0.73$ , $p=$ 2.00\text{\times}{10}^{-04}$$	$r(11)=0.76$ , $p=$ 2.44\text{\times}{10}^{-03}$$	$r(45)=0.61$ , $p=$ 6.51\text{\times}{10}^{-06}$$	$r(34)=0.59$ , $p=$ 1.38\text{\times}{10}^{-04}$$
y-vel-avg	$r(19)=0.73$ , $p=$ 1.65\text{\times}{10}^{-04}$$	$r(11)=0.68$ , $p=$ 1.11\text{\times}{10}^{-02}$$	$r(45)=0.46$ , $p=$ 1.16\text{\times}{10}^{-03}$$	$r(34)=0.66$ , $p=$ 1.15\text{\times}{10}^{-05}$$
pupil-avg	$r(19)=0.72$ , $p=$ 2.19\text{\times}{10}^{-04}$$	$r(11)=0.68$ , $p=$ 9.97\text{\times}{10}^{-03}$$	$r(45)=0.47$ , $p=$ 7.48\text{\times}{10}^{-04}$$	$r(34)=0.48$ , $p=$ 3.27\text{\times}{10}^{-03}$$
x-pos-var	$r(19)=0.53$ , $p=$ 1.29\text{\times}{10}^{-02}$$	$r(11)=0.73$ , $p=$ 4.49\text{\times}{10}^{-03}$$	$r(45)=0.37$ , $p=$ 1.13\text{\times}{10}^{-02}$$	$r(34)=0.56$ , $p=$ 3.99\text{\times}{10}^{-04}$$

Table S4 : Partial correlation between information redundancy and learning index controlling for eye dynamic variables. Each item shows the Spearman correlation between

I_{\rm redundancy}

and learning index for one epoch (column) with one eye dynamic variable (row) as the control variable. In this table, all correlations remain significantly positive.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
Largest PC number	8	7	7	3
Explained variance	98%	97%	92%	59%
Partial correlation	$r(12)=0.55$ $p=$ 4.22\text{\times}{10}^{-02}$$	$\rho(5)=0.84$ $p=$ 1.79\text{\times}{10}^{-02}$$	$r(39)=0.42$ $p=$ 6.94\text{\times}{10}^{-03}$$	$r(32)=0.37$ $p=$ 3.20\text{\times}{10}^{-02}$$

Table S5 : Partial correlation between

I_{\rm redundancy}

and learning index with principal components as control variables. In each column, we report the largest number of principal components (first row) with which the partial correlation between

I_{\rm redundancy}

and learning index (third row) was still significant, and the variance they could explain (second row) for the ten eye dynamic quantities.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
corr (learning index, choice decoding accuracy)	$r(20)=-0.05$ , $p=$ 8.09\text{\times}{10}^{-01}$$	$r(12)=0.42$ , $p=$ 1.32\text{\times}{10}^{-01}$$	$r(46)=0.68$ , $p=$ 3.68\text{\times}{10}^{-07}$$	$r(35)=0.34$ , $p=$ 4.20\text{\times}{10}^{-02}$$
Partial correlation ( $I_{\rm redundancy}$ , learning index $\|$ choice decoding accuracy)	$r(19)=0.74$ , $p=$ 1.08\text{\times}{10}^{-04}$$	$r(11)=0.78$ , $p=$ 1.53\text{\times}{10}^{-03}$$	$r(45)=0.40$ , $p=$ 4.78\text{\times}{10}^{-03}$$	$r(34)=0.69$ , $p=$ 2.79\text{\times}{10}^{-06}$$

Table S6 : Decoding choice with eye dynamic variable. The first row shows Spearman correlations between the learning index and decoding accuracy of choice from eye dynamic variables. The second row shows the Spearman correlation between

I_{\rm redundancy}

and learning index, controlling for the choice decoding accuracy.

	Monkey R, cardinal	Monkey R, oblique	Monkey G, oblique	Monkey G, cardinal
Psychometric threshold	$3.29\pm 1.08\%$	$5.77\pm 1.90\%$	$17.49\pm 5.35\%$	$7.60\pm 1.41\%$
Neurometric threshold	$4.52\pm 1.75\%$	$5.20\pm 1.16\%$	$11.78\pm 8.33\%$	$10.09\pm 2.19\%$
Population N/P Ratio	$1.45\pm 0.45$	$0.90\pm 0.14$	$0.69\pm 0.32$	$1.35\pm 0.32$

Table S7 : Psychometric and populational neurometric threshold late in learning. We report the coherence levels for which the animal reached 75% accuracy for each task (Mean

\pm

s.t.d. across late-learning sessions). The third row: the ratios between neurometric and psychometric thresholds.

Neural Encoding

Neural encoding studies how information is represented in biological neural activity and how models map stimuli or signals into neural codes.

See all 91 papers

May 28, 2026

Supervised Training Rapidly Degrades Early Visual Cortex Alignment Across Biologically Plausible Learning Rules

Random, untrained neural networks consistently match or exceed trained networks in representational similarity to early visual cortex. This puzzling finding challenges the assumption that learning improves brain alignment. We investigate it by tracking representational similarity analysis (RSA) alignment to human fMRI data across training for four learning rules: backpropagation (BP), feedback alignment (FA), predictive coding (PC), and spike-timing-dependent plasticity (STDP). Using 720 object images from the THINGS database and fMRI data from three subjects across six visual ROIs, we measure Spearman correlations between model and brain representational dissimilarity matrices at eight training checkpoints (epochs 0-40). We find that (1) a single epoch of training reduces V1 alignment by 25-90%, depending on the learning rule; (2) backpropagation reduces V1 alignment most severely (delta r = -0.080), while predictive coding and STDP preserve substantially more (delta r ~ -0.04); and (3) a weaker, opposite tendency appears in object-selective cortex (LOC), where BP shows the largest increase in alignment during training, although the absolute change is small. These results suggest that untrained architectures capture low-level visual statistics through inductive biases alone, and that global error signals (BP) reshape early representations more aggressively than local learning rules (PC, STDP), which better preserve brain-like structure.

Nils Leutenegger

Independent Researcher, Switzerland

Neural Representation Geometry Predictive Coding Neural Processes Visual Neuroscience Biologically Plausible Learning Neural Network Training Dynamics

Ask PDF

May 18, 2026

Efficient coding along the visual hierarchy

Biological visual systems learn from limited experience, unlike deep learning models that rely on millions of training images. What learning principles make this possible? We tested whether efficient coding, the idea that neural representations capture the statistical structure of natural inputs, can build a hierarchy of human-aligned visual features from limited data. We developed an unsupervised learning procedure in which each layer of a deep network compresses its inputs onto the dominant modes of variation in natural images, using only local statistics and no labels, tasks, or backpropagation. This unsupervised procedure yields features that progress from edges and colors to textures and shapes. The features of this deep efficient coding model are readily recognized by human observers and are predictive of image-evoked fMRI responses in human visual cortex. Furthermore, a hybrid learning procedure that combines efficient coding with supervised fine-tuning yields better brain alignment in low-data settings and more rapid category learning. These findings suggest that efficient coding may shape representations across the entire visual hierarchy and help explain the data efficiency of biological vision.

Ananya Passi, Brian S. Robinson, Michael F. Bonner

Department of Cognitive Science, Johns Hopkins University, Baltimore, MD, USA · Applied Physics Laboratory, Johns Hopkins University, Baltimore, MD, USA

Unsupervised Learning Neural Processes Hierarchical Representation Learning Unsupervised Representation Learning Visual Neuroscience Neural Encoding+2

Ask PDF

May 30, 2026

Cortex and subcortex play distinct roles over learning when cortical memory is limited

It has been proposed that the brain integrates flexible, computationally expensive cortical processing with simpler, lower-cost subcortical mechanisms to achieve resource-efficient performance greater than that of either system alone. Despite the allure of this perspective, satisfying theoretical frameworks that explore this hypothesis are still limited. We extend existing frameworks in which a model-based module and model-free module learn in tandem by explicitly constraining the memory resources of the model-based module, and investigate the impact of this constraint in a simple decision-making setting. Memory constraints naturally give rise to strategies for allocating memory resources. We evaluate the performance of different strategies in different situations and demonstrate that when the rewarded states change often, it can be advantageous for the model-based module to focus its memory resources not on exploiting the current reward, but on capturing general structure of the environment. This work provides a theoretical foundation for a functional dissociation between cortical and subcortical systems during learning: the cortex supports general structure learning, while subcortical circuits specialize in reward-based learning. We further detail how these hypotheses can be tested on experimental data.

Matthew Farrell, Taro Toyoizumi

Laboratory for Neural Computation and Adaptation RIKEN Center for Brain Science, Wako, Japan · Department of Mathematical Informatics, Graduate School of Information Science and Technology, The University of Tokyo, Tokyo, Japan.

Computational Cognitive Modeling Reinforcement Learning Neural Processes Model-Free RL Model-Based RL

Ask PDF

Abstract

Neural Encoding

Explore similar work

Supervised Training Rapidly Degrades Early Visual Cortex Alignment Across Biologically Plausible Learning Rules

Efficient coding along the visual hierarchy

Cortex and subcortex play distinct roles over learning when cortical memory is limited