Attention and normalization circuits in macaque V1

Abstract

Attention affects neuronal processing and improves behavioural performance. In extrastriate visual cortex these effects have been explained by normalization models, which assume that attention influences the circuit that mediates surround suppression. While normalization models have been able to explain attentional effects, their validity has rarely been tested against alternative models. Here we investigate how attention and surround/mask stimuli affect neuronal firing rates and orientation tuning in macaque V1. Surround/mask stimuli provide an estimate to what extent V1 neurons are affected by normalization, which was compared against effects of spatial top down attention. For some attention/surround effect comparisons, the strength of attentional modulation was correlated with the strength of surround modulation, suggesting that attention and surround/mask stimulation (i.e. normalization) might use a common mechanism. To explore this in detail, we fitted multiplicative and additive models of attention to our data. In one class of models, attention contributed to normalization mechanisms, whereas in a different class of models it did not. Model selection based on Akaike's and on Bayesian information criteria demonstrated that in most cells the effects of attention were best described by models where attention did not contribute to normalization mechanisms. This demonstrates that attentional influences on neuronal responses in primary visual cortex often bypass normalization mechanisms.

Keywords: attention; normalization; orientation tuning; striate cortex; surround suppression.

Figure 1

Task and stimuli used. Animals were required to fixate a fixation point (FP) on the monitor. A cue indicated to the animal where to attend to on the current trial. Then, 900 ms after cue offset two stimuli appeared, one centred on the neuron's RF and one in the opposite hemifield equidistant to the FP. Animals had to detect a subtle change in the middle of the centre bar in the cued location, and ignore changes in the un-cued location. The left timeline shows the conditions when the centre stimulus was surrounded by distractor stimuli (with surround condition). The right timeline shows the condition when the centre bar was presented in isolation (no surround condition). The exact orientation of surrounding bars used on every trial in the experiments is shown here.

Figure 2

(A) Effects of attention and surround stimulation on orientation tuning in an example cell. Filled squares and circles show the mean activity (y-axis) recorded given the stimulus orientation (x-axis), and solid continuous lines show the wrapped Gaussians fitted to these data points. Attention (grey lines) increased the neuronal activity when stimuli were close to the preferred orientation. Surround stimuli reduced the neuronal activity at all orientations (dashed lines). Error bars show SEM. (B) Effects of attention and surround stimulation on orientation tuning on the population of cells. Solid fitted lines show the effects when no surround stimuli were presented, and dashed lines show the effects when a surround stimulus was presented. (C) Normalized population PSTHs when no surround stimulus was presented (solid lines) and when a surround stimulus was presented (dashed lines). Shaded areas show SEM. Time zero corresponds to stimulus onset. Grey lines show the neuronal activity when stimuli were attended to, and black lines when attention was directed into the opposite hemifield.

Figure 3

(A) Suppression index (SI) plotted against the preferred orientation of a given cell. Preferred orientations in the cell sample were relatively uniformly distributed (spread of points along the x-axis in the upper graph). SIs were mostly positive, i.e. in most cells surround stimuli reduced the neuronal activity. (B) Orientation of the bars in the surround. Grey tick marks show the orientation of bars on the inner ring, and black tick marks show the orientation of bars on the outer ring of the surround stimulus. (C) Attentional modulation indices plotted against the preferred orientation of a given cell. Attentional modulation indices were mostly positive, i.e. in most cells attention increased neuronal activity. Neither suppression nor attentional modulation indices showed an obvious relationship to preferred orientation of the neurons, or to the distribution of orientations in the surround stimulus.

Figure 4

Effects of surround stimuli on tuning amplitude, baseline and bandwidth of the wrapped Gaussian fitted to the responses of each cell. Effects are shown separately for the attend away (black) and the attend RF (grey) condition. Centre plots (and associated text: r = correlation coefficient, P = significance of correlation) show whether modulation indices (MIs) were correlated between amplitude, baseline and tuning width. Histograms above and to the side of centre plots show distributions of MIs. Associated text indicates the median of these distributions and whether these were significantly different from 0 (associated P-value).

Figure 5

Correlation of attentional and surround modulation index (based on raw firing rates averaged for the preferred stimulus orientation). Four different comparisons were possible which are shown in A–D, respectively. Specifically, we obtained an attentional MI when no surround stimulus was present and when a surround stimulus was present. Moreover we obtained a surround MI when attention was directed away from the RF and one when it was directed towards the RF. Text insets report the correlation coefficient (r) and the significance of the correlation (P).

Figure 6

Pairwise distribution of Akaike weights for three different models. (A) Distribution of Akaike weights for the additive non-normalizing two-parameter model (x-axis) vs. the additive normalizing two-parameter model. The average weight (±SEM) for both models is indicated by the location of the square and error bars. (B) Distribution of Akaike weights for the additive non-normalizing two-parameter model (x-axis) vs. the multiplicative non- normalizing two-parameter model. The average weight (±SEM) for both models is indicated by the location of the square and error bars. Histograms along the axes show the frequency distributions.

Figure 7

Comparison of measured against predicted responses. (A) Population average of measured responses. Red line shows the mean population response in the attend RF – no surround condition (shaded area shows SEM). The magenta line shows the attend RF – with surround responses, and the blue line shows the attend away – with surround responses. (B) Predicted responses from the multiplicative non-normalizing model (colour coding as in A). (C) Predicted responses from the additive non-normalizing model (colour coding as in A).

Figure 8

Correlation of fitting parameters with modulation indices. (A) Correlation between the attentional and normalization fitting parameter from the multiplicative model (single parameter model) and measured attention and surround modulation indices. (B–E) Correlation between the attentional and normalization fitting parameters from the multiplicative non-normalizing two-parameter model with the measured attention and surround modulation indices. (F–I) Correlation between the fitting parameters from the best fitting additive attention non-normalizing two-parameter model and measured attention and surround modulation indices. The normalization term captures surround effects, while the parameter β captures attention effects. Text insets give correlation coefficients (r, Spearman rank correlation) and significance (P). Grey and black symbols/text delineate different MI conditions as labelled along the x- and y-axes.