Shaobo Guan (Summit Kwan)

The Questions: Our vision begins from the retina, and thus many neurons in the vast expanse of the visual system exhibit retinotopic receptive fields ("retinitopic" means eye-centered, see more at wiki).  Accordingly, objects in the visual field are initially localized based on the retinotopic coordinates.  This coordinate frame works well if our gaze remains still.  However, in reality, our gaze frequently changes location, by means like saccades (a type of fast eye movement, which occurs ~3 times per second, see more at wiki).  Each gaze shift induces a sweeping displacement of the visual image on the retina, resulting in an entirely new retinotopic map of space.  Therefore, transforming visual inputs to a gaze-invariant frame or updating the retinotopic representation properly is essential for localizing a visual stimulus across a gaze shift.  This spatial coding strategy is known to be critical for implementing visual stability and sensory-motor control.

Previous Hypotheses: Two hypotheses have been proposed to solve the spatial coding problem (see here for a comprehensive review from Robert Wurtz, 2008, vision research): i) spatiotopic hypothesis, in which the retinotopic representation is transformed into a spatiotopic one, so that the object location remains invariant across a gaze shift; and ii) retinotopic hypothesis, in which the retinotopic representation updates, or remaps, with a gaze shift, so as to be consistent with the consequence of the gaze shift.  So far, both hypotheses have been supported by monkey electrophysiology experiments.  Here we ask the question: Are these two hypotheses really incompatible with each other or are they merely two sides of the same coin?

A New experimental Finding: In a recent study, Zhang et al. explored the detailed spatial-temporal structure of saccadic remapping.  They found that neurons in monkey posterior parietal cortex elongate their receptive fields along the saccadic trajectory, effectively stretching the receptive fields along the vector of the eye movement (this observation is different from what Sommer and Wurtz reported in frontal lobe, where neurons’ receptive fields “jump” during the remapping).  To us, this new finding is inspirational, because we think such spatial-temporal profile may have betrayed the underlying network dynamics that are critical for the trans-gaze-shift spatial coding.

Our solution: We built a mechanistic neural network model to decipher the strategy of coordinate transformation that could be used in posterior parietal lobe, a cortical area specialized in processing spatial information.  Our key assumption is that neurons preferring the same/similar spatiotopic position are strongly connected.  With this simple assumption, our recurrent basis function network model implements the following experimental findings: i) gaze-dependent gain modulation on the retinotopic visual response, ii) predictive visual remapping right before a saccade, iii) invariance of spatiotopic representation and updating of retinotopic representation across gaze shifts, and iv) elongation of neuronal receptive fields along the trajectory of a saccade.  Our model reconciles the spatiotopic and retinotopic hypotheses, and sheds light on understanding the principles underlying the coordinate transformations in the brain.

Hope this manuscript can be published soon so that you can access the full-text.

Our perception and behavior can be largely regulated by the attention, i.e. 1) to lower threshold (detect the subtle sound that is otherwise ignored), 2) to increase precision (tell the tiny differences when playing Finding the Fault), and 3) to speed up our response (that’s why it’s better to be concentrated during driving).

How attention modulates our brain?  To investigate, we choose to narrow down this huge questing a little bit by asking “How attention regulates visual-oculomotor transformation process”.  This is because the visually guided saccade (saccade: a type of fast eye movement that occurs ~3 times per second in alert human, see wiki link) is a well studied paradigm, which makes it feasible to explore the role that attention plays in this relatively simple process.  To further narrow down this question, we use the reaction time of saccades as a behavioral probe to explore the attention effect, which could be tested by simply doing psychophysical experiments with human subjects.

In our study, we take advantage of two well know phenomena, both of which has been separately studied for ~30 years.  One is Inhibition of Attention Return (IOR, see more at the wiki link), which reports the inhibited “tagging” to previously attended position, reflected as prolonged mean reaction times; the other is bimodal distribution of reaction times, which demonstrates two temporally separated modes of saccade (express vs. regular), reflecting two distinct underlying neural pathways.  Therefore, by combining IOR and bimodal reaction times together, how would we expect attention will affect reaction time distribution?  If attention routes the visual-motor signal transformation between different pathways, the proportion of the two modes will be modulated in the reaction time distribution (routing hypothesis); conversely, if attention modulates the general signal transmission speed, the two modes will be shifted along the time axis (shifting hypothesis).

Following the above rational, we tested human subjects under a novel Cue Gap Paradigm, and recorded their eye movements with an infrared eye tracker.  Our results showed that attention regulates mean reaction time by changing the proportion of the two modes, instead of shifting both modes along the time axis, therefore supports the routing hypothesis.

This work is a straightforward psychophysical study on how attention affects sensory-motor transformation; besides, by combing IOR and bimodal distribution together for the first time, it shed light on both of them and generates some conjectures on the underlying circuits.  My colleagues are now exploring the neural substrates of our conclusion through doing recordings on alert animals.  Hope we can soon substantiate the story.

The manuscript of this psychophysical experiment has been accepted for publication at Journal of Neurophysiology, with the title "Covert attention regulates saccadic reaction time by routing between different visual-oculomotor pathways". You can access it here