Having established that both tuning to both orientation and to fine-scale retinotopy may be re- solved with our available neuroimaging techniques, our experiments follow a similar framework to ask whether attention to an orientation (Chapter 4) or to a restricted region of space (Chap- ter 5) generates attentional modulations that are (A) specifically observed within the known
22 subpopulation of V1 that is tuned toward the attended stimulus, or (B) non-specifically to a larger extent of V1.
We have developed a general framework to measure the specificity of attentional modulations. Briefly, we select a set of stimuli that evoke differential neural responses patterns within V1, and we measure these patterns utilizing high resolution imaging methods. These will be stimuli of different orientations, activating distinct sets of orientation columns (Chapter 4), or stimuli at different monitor locations, activating distinct regions of the V1 retinotopic map (Chapter 5). We then, under the same task context and measurement conditions, cue subjects to attend to one or another of the mapped stimuli. We recover a pattern of attentional modulations evoked by this cue, and we test for a relationship between the activity patterns of the attentional modulation and of the underlying neural tuning functions. The strength of this relationship provides a measurement, to the precision of our imaging techniques, as to how precisely these attentional modulations are targeted to the most well-tuned neural populations within the brain for a given task.
Chapter 4
Functional Magnetic Resonance
Imaging of Attentional
Modulations at the
Near-Columnar Scale
At higher magnetic fields, the hemodynamic signals recovered by BOLD fMRI are increas- ingly sensitive to signals from smaller venules which drain blood from more localized regions of cortex[117]. Leveraging this basic phenomenon, MRI physicists have developed the capability to recover, at ultra-high magnetic field, functional activity from visual cortex with 0.9-1.5 orders of magnitude greater precision than may be derived from conventional fMRI (0.8-1.5 mm isomet- ric voxels versus conventional 3 mm isometric voxels)[118, 119]. While this increase in spatial precision does not yet deliver true resolution of the activity from individual cortical columns, it is close enough to permit the extrapolation of columnar circuitry from individual voxels.
Using this technology, we measured orientation tuning curves from V1 of human volunteers while simultaneously cuing the subjects to attend to one orientation. We found that this atten- tional cue resulted in the direct facilitation of orientation-tuned responses from individual V1 voxels that had an orientation preference aligned toward the cue independent of the location of these voxels within the V1 retinotopic map. This result suggests that attention is in fact selectively targeted to individual neural subpopulations as a function of their ability to encode the attended stimulus.
24
4.1
Introduction
A defining feature of visual attention is its flexibility. Subjects may selectively attend to loca- tions, objects, periods of time and visual features to enhance their perceptual capabilities[120, 121, 122, 46]. Of these, the selection according to location (spatial attention) is the most stud- ied. Numerous studies have demonstrated that when subjects covertly attend to a location, the sensory responses of neurons representing this location are enhanced throughout the visual hierarchy[123, 78].
Studies of single neurons in monkey visual cortex suggest that non-spatial attention is sim- ilarly targeted, such that attention preferentially enhances neurons selective for an attended feature[124] and attentional modulations are strongest during times that the animal is maxi- mally focused[28]. These attentional modulations may be divided into two broad categories: linear, gain-like increases in a neural firing[125]; and more complex non-linear modulations. While a variety of non-linear effects have been reported[126, 81, 83], similar gain modulations have been observed in spatial[78, 2] and featural[127] attention studies. Moreover, computational modeling suggests that some non-linear effects may actually arise from simple gain processes[80]. These findings lead us to hypothesize the existence of a single common mechanism for visual attention: while attending to a stimulus, simple but computationally powerful[85] gain modula- tions are targeted to the neurons and times most appropriate for the task at hand. Testing this theory requires the ability to systematically map the representation of a visual feature represen- tation across an entire visual area to first identify the neural subpopulation best matched to the task and then measure how responses within that subpopulation change with attention and over time. To this end, the encoding of stimulus orientation within primary visual cortex (V1) is ideal. Within V1, a single cortical column contains neurons tuned towards a common orientation[63], and recently developed functional magnetic resonance imaging (fMRI) techniques are capable of measuring orientation tuning at columnar resolutions[103]. Moreover, orientation tuning can be observed even in voxels, which are larger than a cortical column[128]. Such tuning offers an opportunity, for the first time, to map a non-spatial visual representation within a single cortical area and to study how that map is dynamically changed with attention.
To address how representations of visual information are altered by non-spatial attention, we therefore imaged V1 using ultra-high field fMRI (7 T) while subjects performed a periodic non-spatial attention task[128, 101]. We discovered that both orientation tuning and attentional modulations of that tuning, are present within individual voxels. Both the orientation prefer- ences and the response timing of voxels systematically shift towards the featural and temporal foci of attention. These shifts can be explained by a model in which featural and temporal attention cause linear changes in activity preferentially directed during behaviorally appropriate times to neurons with appropriate feature selectively. Our results suggest that representations
at the earliest stages of visual processing can be profoundly altered by cognitive influences and that all forms of attention may act by common mechanisms to selectively enhance behaviorally relevant sensory representations throughout cerebral cortex.