Optimal Deployment of Attentional Gain during Fine Discriminations

Abstract

Most models assume that top-down attention enhances the gain of sensory neurons tuned to behaviorally-relevant stimuli (on-target gain). However, theoretical work suggests that when targets and distracters are highly similar, attention should enhance the gain of neurons that are tuned away from the target, because these neurons better discriminate neighboring features (off-target gain). While it is established that off-target neurons support difficult fine discriminations, it is unclear if top-down attentional gain can be optimally applied to informative off-target sensory neurons or if gain is always applied to on-target neurons, irrespective of task demands. To test the optimality of attentional gain in human visual cortex, we used fMRI and an encoding model to estimate the response profile across a set of hypothetical orientation-selective channels during a difficult discrimination task. The results suggest that top-down attention can adaptively modulate off-target neural populations, but only when the discriminanda are precisely specified in advance. Furthermore, logistic regression revealed that activation levels in off-target orientation channels predicted behavioral accuracy on a trial-by-trial basis. Overall, these data suggest that attention does not always increase the gain of sensory-evoked responses, but instead may bias population response profiles in an optimal manner that respects both the tuning properties of sensory neurons and the physical characteristics of the stimulus array.

Reconstructing representations of dynamic visual objects in early visual cortex

“…Ten tuning runs comprised the dataset for model estimation (m × n matrix B 1 ). To maximize the signal-to-noise ratio of channel responses, we selected the top 50% of voxels within each ROI that could best discriminate the different orientations in the tuning runs (the top 50% of voxels with highest F statistic in ANOVA of response amplitudes in tuning runs) (31).…”

“…To build an encoding model of orientation (29,31), each subject completed tuning runs in which the subject viewed whole-field gratings in eight different orientations (SI Materials and Methods). Using these data, we generated encoding models for hypothetical orientation channels based on the responses to each orientation for each voxel (validated with a leave-one-run-out procedure; Fig.…”

Using Human Neuroimaging to Examine Top-down Modulation of Visual Perception

“…We first review observation-based analysis approaches that focus on decoding stimulus features [8][9][10][11][12][13] and that seek to identify independent variables that best predict brain activity [14][15][16]. Then, we contrast these approaches with complementary representation-based methods that attempt to directly reconstruct stimulus features based on patterns of neural responses [17][18][19][20][21][22][23][24][25][26][27][28]). In all cases, we place particular emphasis on how these different methods can be used to test formal models of top-down cognitive control.…”

“…First, perception and behavior are not typically linked to the response properties of single neurons but are instead thought to be linked more closely with the joint activity of millions of neurons that form population codes representing everything from basic sensory features to complex motor plans [35][36][37][38]. Recently developed methods allow BOLD and EEG measures to assess these population responses with increasingly high precision [17][18][19][20][21][22][23][24][25][26][27][28][39][40][41][42][43][44][45][46][47][48][49][50], and in turn these responses are likely to be much more closely coupled with perception and behavior compared to isolated single-unit spike rates. Second, these imaging tools can be used to assess changes in neural activity associated with nuanced and complex cognitive tasks that human subjects can master in a matter of minutes but that non-human primates would be unable to easily perform.…”

“…3), we here focus on the applications of univariate and multivariate encoding models and their application to identifying the sensitivity profiles of individual voxels to visual [42,44,48,54,55] and semantic [43,45,46] features of the stimulus, and how those sensitivity profiles change as a function of task demands [23,47,[54][55][56]]-all observation-based endeavors. We also will address the novel procedure of "inverting" encoding models to reconstruct region-wide representations of stimulus features given patterns of neural activity [17][18][19][20][21][22][23][24][25][26][27][28]-early investigations using the representation-based approach.…”