Adaptive gain modulation in V1 explains contextual modifications during bisection learning

Abstract

The neuronal processing of visual stimuli in primary visual cortex (V1) can be modified by perceptual training. Training in bisection discrimination, for instance, changes the contextual interactions in V1 elicited by parallel lines. Before training, two parallel lines inhibit their individual V1-responses. After bisection training, inhibition turns into non-symmetric excitation while performing the bisection task. Yet, the receptive field of the V1 neurons evaluated by a single line does not change during task performance. We present a model of recurrent processing in V1 where the neuronal gain can be modulated by a global attentional signal. Perceptual learning mainly consists in strengthening this attentional signal, leading to a more effective gain modulation. The model reproduces both the psychophysical results on bisection learning and the modified contextual interactions observed in V1 during task performance. It makes several predictions, for instance that imagery training should improve the performance, or that a slight stimulus wiggling can strongly affect the representation in V1 while performing the task. We conclude that strengthening a top-down induced gain increase can explain perceptual learning, and that this top-down signal can modify lateral interactions within V1, without significantly changing the classical receptive field of V1 neurons.

Figure 1. Bisection task and model network.

(A) A bisection stimulus consists of three vertical lines, with the middle line slightly displaced from the center. The subject has to indicate whether this middle line is displaced toward the left or the right line. (B) Possible embedding of the model into a V1 circuitry. (Only a subset of the otherwise mirror-symmetric connectivity pattern in the V1-box is rendered, and to show the continuation, the initial segments of some connection lines are also drawn). Each of the bisection lines is activating (via non-modeled L4 neurons) a single L5 pyramidal neuron while projecting through a Gaussian fan out to the L2/3 pyramidal neurons. Both pyramidal layers project to a binary decision unit in a higher cortical area. The gain of the L2/3 pyramidal neurons is modulated by top-down input from a task population. L2/3 neurons are recurrently connected both through direct excitation and via a global inhibitory neuron. (C) The decision unit sums up and thresholds the weighted firing rates of the noisy pyramidal neurons. Learning consist in modifying these readout weights, as well as in a modification of the top-down input strength.

Figure 2. The top-down induced gain increase of the L2/3 neurons provokes symmetry breaking in the recurrent network and the resulting competition improves the signal-to-noise ratio underlying the bisection decisions.

(A) and (B): Network activities for two mirror symmetric bisection stimuli after learning. (i) Each line in the bisection stimulus activates a neuron in L5 at the corresponding position (dashed line indicates the stimulus center). (ii) The feedforward input to the L2/3 pyramidal neurons is locally spread. (iii) Local recurrent excitation and global inhibition competitively suppresses L2/3 pyramidal neurons receiving weak input, leading to a lateralization of the activity to the side of the middle line (A: left; B: right). An additional top-down gain increase enhances this lateralization (black versus grey lines). Deviations from mirror symmetry in the responses are due to a stochastic modulation of the lateral connectivity in L2/3. (iv) The input to the decision unit, , is a weighted sum of the noisy L2/3 and L5 activities without (grey) and with (black) top-down input, upon which the decision ‘left’ or ‘right’ is made by thresholding at . The weak gain increase (by a factor of ) dramatically increases the signal (by a factor of and , respectively). The plots show averaged activities over runs with the same stimulus configurations.

Figure 3. Performance and evolution of the readout weights during bisection training.

(A) Fraction of erroneous network decisions against training week, with a ‘week’ consisting of the presentation of bisection stimuli of fixed outer-line-distance (‘width’) but with randomized positions. Upon each stimulus presentation, the readout weights from the L5 and the L2/3 pyramidal neurons to the decision unit were changed according to an error correcting learning rule. A top-down induced gain increase in the L2/3 pyramidal neurons reduces the error level (grey: gain factor ; black: gain factor ). Hence, a substantial improvement in performance is achieved if learning simultaneously increases the top-down input strength, leading to a learning curve which interpolates between the two curves (dashed line). The fast initial learning progress arises from adapting the readout connections to the decision unit. (B) Learning curve for a monkey performing the bisection task (adapted from [10]). (C) Synaptic weights from L2/3 pyramidal neurons to the decision unit before (circles) and after learning with (black) and without gain increase (grey). The dotted line indicates the universal weight distribution inferred in the theoretical argument. (D) Same as in C, but for synaptic weights from L5 pyramidal neurons to the decision unit. Error bars represent standard error of the mean (using learning runs).

Figure 4. Learning-induced gain modulation in L2/3 pyramidal neurons qualitatively changes the local interactions.

(A) In the experiment , monkeys were performing either a fixation task (top) or a bisection task (bottom) while the activity of a supra-granular V1 neuron was recorded in response to a two line stimulus in a side-by-side configuration. One of the two lines is centered in the receptive field (sketched by the square) of the recorded neuron. (B) Activity of the corresponding model L2/3 pyramidal neuron mimicking the recorded supra-granular neuron for different positions of the flanking line, with individual curves normalized by the activity with flanking line at . Top: During pure fixation or before training (modeled by a non-modulated circuitry, gain ), the response of the central neuron is suppressed by the flanking line via global inhibition. Bottom: When performing the bisection task at a nearby location in the trained hemisphere (modeled by a top-down induced gain increase of the L2/3 pyramidal neurons from to ) the lateral suppression turns into strong excitation at random positions due to the enhanced competition within the stochastically modulated network. (C) Modulation indices for the ‘bisection task’ (gain ) versus ‘fixation task’ (gain ). The modulation index is defined as the normalized difference between the maximal and minimal response of the recorded L2/3 pyramidal neuron, each evaluated for the different positions of the flanking line (as represented in B, see Materials and Methods). Evaluation for neurons in stochastic network configurations shows that the modulation index under the bisection condition is significantly larger than under the fixation condition ( for paired t-test with ), as it is also observed in the experiment ( with , , ).

Figure 5. Largely task-independent receptive field of L2/3 neurons.

(A) Averaged normalized responses of one typical L2/3 model pyramidal neuron to a single line placed at different positions for the un-modulated network (‘fixation task’, gain , dashed line) and with a top-down induced gain increase of the L2/3 pyramidal neurons (‘bisection task’, gain , solid line). Grey lines show Gaussian fits (with and for the fixation and the bisection task, respectively). Error bars arise from the stochasticity in the top-down induced gain modulation ( line presentations at each position with fixed network configuration). (B) Histogram of the differences in the receptive field (RF) size of model pyramidal neurons under bisection versus fixation conditions. For comparison with the experiment where the same number of neurons were recorded from different positions and animals, we extracted the model neurons from different network configurations and determined the receptive field as in A. The difference in the receptive field size was not significant ( in the t-test with ), in agreement with the experimental findings (, with , , ). However, increasing the number of sample neurons may turn a non-significant into a significant result, and for the model this is in fact the case, with RF size during the bisection task becoming significantly (in terms of the t-test) smaller by than without performing this task.

Figure 6. Training under stimulus roving and transfer to untrained stimulus widths.

(A) Fraction of incorrect network decisions for the combined training with two stimulus widths ( and ) which were randomly interleaved (‘roving’). In agreement with recent findings – but unlike previous predictions , – learning under stimulus roving is impaired, although still possible (the final fraction of incorrect responses, being for stimulus roving, is reduced for individual training of the bisection width 5 and 9 to and , respectively). Note that the post-training test shows a better performance for the interpolated width which was itself not trained. (B) Learning curve for bisection stimuli of width (line), with pre- and post-learning tests for the untrained stimulus widths and . A learning transfer of roughly from the trained to the two untrained widths is predicted by the model. Error bars represent the standard error of the mean evaluated for runs.

Figure 7. Steady state activity of the continuously distributed L2/3 neurons (Eq. 11, solid lines).

Feedforward input (dashed lines) and bisection stimulus with lines positions at , and , are the same as in Figure 2B. The width of the recurrent projections () varies for the three sub panels: (A) , (B) and (C) . Symmetry breaking is strongest if is roughly half the bisection width (B, corresponding to the parameter choice in the discrete simulations). For smaller and larger (A and C), the activity to the left bisection line is not fully suppressed and the distribution is less asymmetric (as expressed by a smaller first order moment , taking on values , and from left to right).

Figure 8. Symmetry breaking as a function of the recurrent projection width ().

The symmetry breaking index describes the shift in the center of gravity of the L2/3 steady state activity when displacing the middle bisection line away from the bisection center (Eq. 14). Parameter values: , and the same values , , and as in the other simulations. The maximum of is at , confirming that (for a nonlinear suppression parameter between and roughly ) the optimal recurrent projection width () is in the range of the half width of the bisection stimulus ().