Linking behavior to different measures of cortical activity

Time/Room: Friday, May 15, 2015, 5:00 – 7:00 pm, Talk Room 1
Organizer(s): Justin Gardner1, John Serences2, Franco Pestilli3; 1Stanford University, 2UC San Diego, 3Indiana University
Presenters: Justin Gardner, John Serences, Eyal Seidemann, Aniruddha Das, Farran Briggs, Geoffrey Boynton

< Back to 2015 Symposia

Symposium Description

A plethora of tools are available for visual neuroscientists to study brain activity across different spatiotemporal scales and the BRAIN initiative offers the promise of more. Pipettes and electrodes measure microscopic activity at channels, synapses and single-units. Multi-electrode arrays, calcium imaging, voltage-sensitive dyes and intrinsic imaging measure mesoscale population activity. Human cortical areas can be mapped using fMRI, ECoG and EEG. In principle, the multiplicity of technologies offers unprecedented possibilities to gain information at complementary spatiotemporal scales. Leveraging knowledge across measurement modalities and species is essential for understanding the human brain where the vast majority of what we know comes from non-invasive measurements of brain activity and behavior. Despite the potential for convergence, different methodologies also produce results that appear superficially inconsistent, leading to categorically distinct models of cortical computation subserving vision and cognition. Visual spatial attention provides an excellent case study. A great deal of behavioral work in humans has established that reaction times and discrimination thresholds can be improved with prior spatial information. Measurements of brain activity using very similar protocols have been made using metrics ranging from single-unit responses to functional imaging in both animals and humans. Despite this wealth of potentially complementary data, general consensus has yet to be achieved. Effects of attention on basic visual responses, such as contrast-response, have yielded different conclusions in and across measurements from fMRI (Buracas and Boynton, 2007; Murray, 2008; Pestilli et al, 2011), voltage sensitive dye imaging (Chen and Seidemann, 2012) and single-units and EEG (McAdams and Maunsell, 1999; Williford and Maunsell, 2006; Cohen and Maunsell 2012; Di Russo et al., 2001; Itthipuripat et al., 2014; Kim et al., 2007; Lauritzen et al., 2010; Wang and Wade, 2011). Task-related responses measured with optical imaging (Sirotin and Das, 2009; Cardoso et al., 2012) also suggest some discrepancy across measurements. These disparate results lead to different models that relate neural mechanisms of attention with behavior (e.g. Pestilli et al, 2011; Itthipuripat et al., 2014). Moreover, some attention effects like reduction in neural variance and pairwise correlations (Cohen and Maunsell, 2009; Herrero et al., 2013; Mitchell et al., 2007; 2009; Niebergall et al., 2011), as well as changes in synaptic efficacy (Briggs et al., 2013) can not even be assessed across all measurements. Rather than considering one specific measurement as privileged, providing ground truth, we propose striving for synthesis and explains the totality of evidence. Theoretic modeling (e.g. Reynolds and Heeger, 2009) provides frameworks that offer the potential for reconciling across measurements (Hara et al., 2014). This symposium is aimed at bringing together people using different spatiotemporal scales of measurements with an eye towards synthesizing disparate sources of knowledge about neural mechanisms for visual attention and their role in predicting behavior. Speakers are encouraged to present results from a perspective that allows direct comparison with other measurements, and critically evaluate whether and why there may be discrepancies. Importantly, the discrepancies observed using these different measures can either lead to very different models of basic neural mechanisms or can be used to mutually constrain models linking neural activity to behavior.


Linking brain activity to visual attentional behavior considering multiple spatial-scales of measurement

Speaker: Justin Gardner; Department of Psychology, Stanford University
Authors: Franco Pestilli; Department of Psychological and Brain Sciences, Program in Neuroscience, Indiana University

Understanding the human neural mechanisms that underly behavioral enhancement due to visual spatial attention requires synthesis of knowledge gained across many different spatial scales of measurement and species. Our lab has focused on the measurement of contrast-response and how it changes with attention in humans. Contrast is a key visual variable in that it controls visibility and measurements from single-units to optical-imaging to fMRI find general consistency in that cortical visual areas respond in monotonically increasing functions to increases in contrast. Building on this commonality across multiple spatial-scales of measurement, we have implemented computational models that predict behavioral performance enhancement from fMRI measurements of contrast-response, in which we tested various linking hypotheses, from sensory enhancement, noise reduction to efficient selection. Our analysis of the human data using fMRI suggested a prominent role for efficient selection in determining behavior. Our work is heavily informed by the physiology literature particularly because some properties of neural response, such as efficiency of synaptic transmission or correlation of activity are difficult if not impossible to determine in humans. Nonetheless, discrepancies across measurements suggests potential difficulties of interpretation of results from any single measurement modality. We will discuss our efforts to address these potential discrepancies by adapting computational models used to explain disparate effects across different single-unit studies to larger spatial-scale population measures such as fMRI.

EEG and fMRI provide different insights into the link between attention and behavior in human visual cortex

Speaker: John Serences; Neurosciences Graduate Program and Psychology Department, University of California, San Diego
Authors: Sirawaj Itthipuripat1, Thomas Sprague1, Edward F Ester2, Sean Deering2; 1Neurosciences Graduate Program 2Psychology Department, University of California, San Diego

A fMRI study by Pestilli et al. (2011) established a method for modeling links between attention-related changes in BOLD activation in visual cortex and changes in behavior. The study found that models based on sensory gain and noise reduction could not explain the relationship between attention-related changes in behavior and attention-related additive shifts of the BOLD contrast-response function (CRF). However, a model based on efficient post-sensory read-out successfully linked BOLD modulations and behavior. We performed a similar study but used EEG instead of fMRI as a measure of neural activity in visual cortex (Itthipuripat et al., 2014). Instead of additive shifts in the BOLD response, attention induced a temporally early multiplicative gain of visually evoked potentials over occipital electrodes, and a model based on sensory gain sufficiently linked attention-induced changes in EEG responses and behavior, without the need to incorporate efficient read-out. We also observed differences between attention-induced changes in EEG-based CRFs (multiplicative gain) and fMRI-based CRFs (additive shift) within the same group of subjects who performed an identical spatial attention task. These results suggest that attentional modulation of EEG responses interacts with the magnitude of sensory-evoked responses, whereas attentional modulation of fMRI signals is largely stimulus-independent. This raises the intriguing possibility that EEG and fMRI signals provide complementary insights into cortical information processing, and that these complementary signals may help to better constrain quantitative models that link neural activity and behavior.

Attentional modulations of sub- and supra-threshold neural population responses in primate V1

Speaker: Eyal Seidemann; Department of Psychology and Center for Perceptual Systems The University of Texas at Austin

Voltage-sensitive dye imaging (VSDI) measures local changes in pooled membrane potentials, simultaneously from dozens of square millimeters of cortex, with millisecond temporal resolution and spatial resolution sufficient to resolve cortical orientation columns. To better understand the quantitative relationship between the VSDI signal and spiking activity of a local neural population, we compared visual responses measured from V1 of behaving monkeys using VSDI and single-unit electrophysiology. We found large and systematic differences between response properties obtained with these two techniques. We then used these results to develop a simple computational model of the quantitative relationship between the average VSDI signal and local spiking activity. In this talk I will describe the model and demonstrate how it can be used to interpret top-down attentional modulations observed using VSDI in macaque V1.

Task-related Responses in Intrinsic-Signal Optical Imaging

Speaker: Aniruddha Das; Department of Neuroscience, Psychiatry, and Biomedical Engineering, Columbia University
Authors: Cardoso, M.1,2, Lima, B.2, Sirotin, Y.2; 1Champalimaud Neuroscience Program (CNP), Lisbon, Portugal; 2Department of Neuroscience, Columbia University, New York, NY

There is a growing appreciation of the importance of endogenous, task-related processes such as attention and arousal even at the earliest stages of sensory processing. By combining intrinsic-signal optical imaging with simultaneous electrode recordings we earlier demonstrated a particular task-related response – distinct from stimulus-evoked responses – in primary visual cortex (V1) of macaque monkeys engaged in visual tasks. The task-related response powerfully reflects behavioral correlates of the task, independent of visual stimulation; it entrains to task timing, increasing progressively in amplitude and duration with temporal anticipation; and it correlates with both task-related rewards, and performance. Notably, however, the effect of the task-related response on stimulus-evoked responses – such as the contrast response function (CRF) – remains an open question. For tasks that are stereotyped and independent of visual stimulation, the task- and stimulus-related responses are linearly separable: the task-related component can be subtracted away leaving an imaged contrast response function that is robustly linear with stimulus-evoked spiking. When the task-related response is modified – such as, by increasing the reward size – the effect is largely additive: the baseline imaging response increases, without, to first order, changing the CRF of the stimulus-evoked component. Thus the important question remains: are there other reliable measures of changes in neural activity, such as changes in signal or noise correlation, rather than local spike rate or LFP magnitude, that can better characterize the task-related response?

Attention and neuronal circuits

Speaker: Farran Briggs; Geisel School of Medicine at Dartmouth College

Visual attention has a profound impact on perception, however we currently lack a neurobiological definition of attention. In other words, we lack an understanding of the cellular and circuit mechanisms underlying attentional modulation of neuronal activity in the brain. The main objective of my research is to understand how visual spatial attention alters the way in which neurons communicate with one another. Previously, my colleagues and I demonstrated that attention enhances the efficacy of signal transmission in the geniculocortical circuit. Through this work, we suggest that the mechanisms underlying attentional modulation of neuronal activity involve enhancement of signal transmission in neuronal circuits and increasing the signal-to-noise ratio of information transmitted in these circuits. Results from my lab indicate that these mechanisms can explain attentional modulations in firing rate observed in primary visual cortical neurons. Our current research focuses on understanding the rules governing attentional modulation of different functional circuits in the visual cortex. Preliminary results suggest that attention differentially regulates the activity of neuronal circuits dependent on the types of information conveyed within those circuits. Overall, our results support a mechanistic definition of attention as a process that alters the dynamics of communication in specific neuronal circuits. I believe this circuit-level understanding of how attention alters neuronal activity is required in order to develop more targeted and effective treatments for attention deficits.

A comparison of electrophysiology and fMRI signals in area V1

Speaker: Geoffrey Boynton; University of Washington, Seattle, WA

fMRI measures in area V1 typically show remarkable consistency with what is expected from monkey electrophysiology studies. However, discrepancies between fMRI and electrophysiology appear for non-stimulus driven factors such as attention and visual awareness. I will discuss possible explanations for these discrepancies, including the role of LFP’s in the hemodynamic coupling process, the effects of feedback and timing, and the overall sensitivity of the BOLD signal.

< Back to 2015 Symposia