Normalization governs attentional modulation within human visual cortex

Goal-directed action preparation in humans entails a mixture of corticospinal neural computations

The seemingly effortless ability of humans to transition from thinking about actions to initiating them relies on sculpting corticospinal output from primary motor cortex. This study tested whether canonical additive and multiplicative neural computations, well-described in sensory systems, generalize to the corticospinal pathway during human action preparation. We used non-invasive brain stimulation to measure corticospinal input-output across varying action preparation contexts during instructed-delay finger response tasks. Goal-directed action preparation was marked by increased multiplicative gain of corticospinal projections to task-relevant muscles and additive suppression of corticospinal projections to non-selected and task-irrelevant muscles. Individuals who modulated corticospinal gain to a greater extent were faster to initiate prepared responses. Our findings provide physiological evidence of combined additive suppression and gain modulation in the human motor system. We propose these computations support action preparation by enhancing the contrast between selected motor representations and surrounding background activity to facilitate response selection and execution.

Perceptual Resolution of Ambiguity: Can Tuned, Divisive Normalization Account for both Interocular Similarity Grouping and Difference Enhancement

Our visual system usually provides a unique and functional representation of the external world. At times, however, the visual system has more than one compelling interpretation of the same retinal stimulus; in this case, neural populations compete for perceptual dominance to resolve ambiguity. Spatial and temporal context can guide perceptual experience. Recent evidence shows that ambiguous retinal stimuli are sometimes resolved by enhancing either similarity or differences among multiple percepts. Divisive normalization is a canonical neural computation that enables context-dependent sensory processing by attenuating a neuron's response by other neurons. Experiments here show that divisive normalization can account for perceptual representations of either similarity enhancement (so-called grouping) or difference enhancement, offering a unified framework for opposite perceptual outcomes.

Spatial attention alters visual cortical representation during target anticipation

Attention enables us to efficiently and flexibly interact with the environment by prioritizing some image features in preparation for responding to a stimulus. Using a concurrent psychophysics- fMRI experiment, we investigated how covert spatial attention affects responses in human visual cortex prior to target onset, and how it affects subsequent behavioral performance. Performance improved at cued locations and worsened at uncued locations, relative to distributed attention, demonstrating a selective tradeoff in processing. Pre-target BOLD responses in cortical visual field maps changed in two ways: First, there was a stimulus-independent baseline shift, positive in map locations near the cued location and negative elsewhere, paralleling the behavioral results. Second, population receptive field centers shifted toward the attended location. Both effects increased in higher visual areas. Together, the results show that spatial attention has large effects on visual cortex prior to target appearance, altering neural response properties throughout and across multiple visual field maps.

Normalization by orientation-tuned surround in human V1-V3

An influential account of neuronal responses in primary visual cortex is the normalized energy model. This model is often implemented as a multi-stage computation. The first stage is linear filtering. The second stage is the extraction of contrast energy, whereby a complex cell computes the squared and summed outputs of a pair of the linear filters in quadrature phase. The third stage is normalization, in which a local population of complex cells mutually inhibit one another. Because the population includes cells tuned to a range of orientations and spatial frequencies, the result is that the responses are effectively normalized by the local stimulus contrast. Here, using evidence from human functional MRI, we show that the classical model fails to account for the relative responses to two classes of stimuli: straight, parallel, band-passed contours (gratings), and curved, band-passed contours (snakes). The snakes elicit fMRI responses that are about twice as large as the gratings, yet a traditional divisive normalization model predicts responses that are about the same. Motivated by these observations and others from the literature, we implement a divisive normalization model in which cells matched in orientation tuning ("tuned normalization") preferentially inhibit each other. We first show that this model accounts for differential responses to these two classes of stimuli. We then show that the model successfully generalizes to other band-pass textures, both in V1 and in extrastriate cortex (V2 and V3). We conclude that even in primary visual cortex, complex features of images such as the degree of heterogeneity, can have large effects on neural responses.

Attention preserves the selectivity of feature-tuned normalization

Attention and divisive normalization both contribute to making visual processing more efficient. Attention selectively increases the neural gain of relevant information in the early visual cortex, resulting in stronger perceived salience for attended regions or features. Divisive normalization improves processing efficiency by suppressing responses to homogeneous inputs and highlighting salient boundaries, facilitating sparse coding of inputs. Theoretical and empirical research suggest a tight link between attention and normalization, wherein attending to a stimulus results in a release from normalization, thereby allowing for an increase in neural response gain. In the present study, we address whether attention alters the qualitative properties of normalization. Specifically, we examine how attention influences the feature-tuned nature of normalization, whereby suppression is stronger between visual stimuli whose orientation contents are similar, and weaker when the orientations are different. Ten human observers viewed stimuli that varied in orientation content while we acquired fMRI BOLD responses under two attentional states: attending toward or attending away from the stimulus. Our results indicate that attention does not alter the specificity of feature-tuned normalization. Instead, attention seems to enhance visuocortical responses evenly, regardless of the degree of orientation similarity within the stimulus. Since visuocortical responses exhibit adaptation to statistical regularities in natural scenes, we conclude that while attention can selectively increase the gain of responses to attended items, it does not appear to alter the ecologically relevant correspondence between orientation differences and strength of tuned normalization.NEW & NOTEWORTHY The magnitude of visuocortical BOLD responses scales with orientation differences in visual stimuli, with the strongest response suppression for collinear stimuli and least suppression for orthogonal, in a way that appears to match natural scene statistics. We examined the effects of attention on this feature-tuned property of suppression and found that while attending to a stimulus increases the overall gain of visuocortical responses, the qualitative properties of feature-tuning remain unchanged, suggesting attention preserves tuned normalization properties.

Divisive normalization is an efficient code for multivariate Pareto-distributed environments

Divisive normalization is a canonical computation in the brain, observed across neural systems, that is often considered to be an implementation of the efficient coding principle. We provide a theoretical result that makes the conditions under which divisive normalization is an efficient code analytically precise: We show that, in a low-noise regime, encoding an n-dimensional stimulus via divisive normalization is efficient if and only if its prevalence in the environment is described by a multivariate Pareto distribution. We generalize this multivariate analog of histogram equalization to allow for arbitrary metabolic costs of the representation, and show how different assumptions on costs are associated with different shapes of the distributions that divisive normalization efficiently encodes. Our result suggests that divisive normalization may have evolved to efficiently represent stimuli with Pareto distributions. We demonstrate that this efficiently encoded distribution is consistent with stylized features of naturalistic stimulus distributions such as their characteristic conditional variance dependence, and we provide empirical evidence suggesting that it may capture the statistics of filter responses to naturalistic images. Our theoretical finding also yields empirically testable predictions across sensory domains on how the divisive normalization parameters should be tuned to features of the input distribution.

Cascaded normalizations for spatial integration in the primary visual cortex of primates

Spatial integration of visual information is an important function in the brain. However, neural computation for spatial integration in the visual cortex remains unclear. In this study, we recorded laminar responses in V1 of awake monkeys driven by visual stimuli with grating patches and annuli of different sizes. We find three important response properties related to spatial integration that are significantly different between input and output layers: neurons in output layers have stronger surround suppression, smaller receptive field (RF), and higher sensitivity to grating annuli partially covering their RFs. These interlaminar differences can be explained by a descriptive model composed of two global divisions (normalization) and a local subtraction. Our results suggest suppressions with cascaded normalizations (CNs) are essential for spatial integration and laminar processing in the visual cortex. Interestingly, the features of spatial integration in convolutional neural networks, especially in lower layers, are different from our findings in V1.

The specificity of orientation-tuned normalization within human early visual cortex

Normalization within visual cortex is modulated by contextual influences; stimuli sharing similar features suppress each other more than dissimilar stimuli. This feature-tuned component of suppression depends on multiple factors, including the orientation content of stimuli. Indeed, pairs of stimuli arranged in a center-surround configuration attenuate each other's response to a greater degree when oriented collinearly than when oriented orthogonally. Although numerous studies have examined the nature of surround suppression at these two extremes, far less is known about how the strength of tuned normalization varies as a function of continuous changes in orientation similarity, particularly in humans. In this study, we used functional magnetic resonance imaging (fMRI) to examine the bandwidth of orientation-tuned suppression within human visual cortex. Blood-oxygen-level-dependent (BOLD) responses were acquired as participants viewed a full-field circular stimulus composed of wedges of orientation-bandpass filtered noise. This stimulus configuration allowed us to parametrically vary orientation differences between neighboring wedges in gradual steps between collinear and orthogonal. We found the greatest suppression for collinearly arranged stimuli with a gradual increase in BOLD response as the orientation content became more dissimilar. We quantified the tuning width of orientation-tuned suppression, finding that the voxel-wise bandwidth of orientation tuned normalization was between 20° and 30°, and did not differ substantially between early visual areas. Voxel-wise analyses revealed that suppression width covaried with retinotopic preference, with the tightest bandwidths at outer eccentricities. Having an estimate of orientation-tuned suppression bandwidth can serve to constrain models of tuned normalization, establishing the precise degree to which suppression strength depends on similarity between visual stimulus components.NEW & NOTEWORTHY Neurons in the early visual cortex are subject to divisive normalization, but the feature-tuning aspect of this computation remains understudied, particularly in humans. We investigated orientation tuning of normalization in human early visual cortex using fMRI and estimated the bandwidth of the tuned normalization function across observers. Our findings provide a characterization of tuned normalization in early visual cortex that could help constrain models of divisive normalization in vision.

Temporal attention selectively enhances target features

Temporal attention, the allocation of attention to a moment in time, improves perception. Here, we examined the computational mechanism by which temporal attention improves perception, under a divisive normalization framework. Under this framework, attention can improve perception of a target signal in three ways: stimulus enhancement (increasing gain across all sensory channels), signal enhancement (selectively increasing gain in channels that encode the target stimulus), or external noise exclusion (reducing the gain in channels that encode irrelevant features). These mechanisms make diverging predictions when a target is embedded in varying levels of noise: stimulus enhancement improves performance only when noise is low, signal enhancement improves performance at all noise intensities, and external noise exclusion improves performance only when noise is high. To date, temporal attention studies have used noise-free displays. Therefore, it is unclear whether temporal attention acts via stimulus enhancement (amplifying both target features and noise) or signal enhancement (selectively amplifying target features) because both mechanisms predict improved performance in the absence of noise. To tease these mechanisms apart, we manipulated temporal attention using an auditory cue while parametrically varying external noise in a fine-orientation discrimination task. Temporal attention improved perceptual thresholds across all noise levels. Formal model comparisons revealed that this cuing effect was best accounted for by a combination of signal enhancement and stimulus enhancement, suggesting that temporal attention improves perceptual performance, in part, by selectively increasing gain for target features.

Differential impact of endogenous and exogenous attention on activity in human visual cortex

How do endogenous (voluntary) and exogenous (involuntary) attention modulate activity in visual cortex? Using ROI-based fMRI analysis, we measured fMRI activity for valid and invalid trials (target at cued/un-cued location, respectively), pre- or post-cueing endogenous or exogenous attention, while participants performed the same orientation discrimination task. We found stronger modulation in contralateral than ipsilateral visual regions, and higher activity in valid- than invalid-trials. For endogenous attention, modulation of stimulus-evoked activity due to a pre-cue increased along the visual hierarchy, but was constant due to a post-cue. For exogenous attention, modulation of stimulus-evoked activity due to a pre-cue was constant along the visual hierarchy, but was not modulated due to a post-cue. These findings reveal that endogenous and exogenous attention distinctly modulate activity in visuo-occipital areas during orienting and reorienting; endogenous attention facilitates both the encoding and the readout of visual information whereas exogenous attention only facilitates the encoding of information.