Update: I am setting up a cognitive neuroscience lab at the Centre for Neuroscience, IISc, Bangalore. I am looking for project associates and students with a background/interest in computer programming, psychology, functional imaging or behavioral neuroscience. Interested in joining the team? Email me!
Nov 24: Officially joined as Assistant Professor at the Centre for Neuroscience, IISc!
Nov 18: Review article on midbrain gamma oscillations is to be published in the Current Opinion in Neurobiology!
Aug 21: Paper on distinguishing bias from sensitivity effects is now online in the Journal of Vision. Read article
Jun 02: Primer on selective attention in birds is published in Current Biology. Read article
May 13: Paper on visuospatial selective attention in birds is published in PNAS. Read article
How does our brain enable us to pay attention selectively to some things, and to ignore others?
What happens in the brain when we make important decisions? My research focuses on understanding the
neural basis of cognitive phenomena such as selective attention and decision making.
The overarching goal is to develop a unified theoretical framework that describes how cognitive phenomena emerge from
neural computations by a systematic analysis of neurons and behavior.
Attention is central to cognition. Despite decades of research on attention, an important question remains unresolved: How do neural computations at various spatial and temporal scales give rise to this emergent behavioral phenomenon? My research aims to forge this missing link at interface of neurons and behavior.
Such an endeavour demands an interdisciplinary approach, a fine interplay between experiments and theory. My research employs many experimental techniques to acquire knowledge at diverse levels: from single neurons and neural circuits, to the behavior of the entire organism. Theoretical and computational approaches have enabled integrating this knowledge into normative, testable models.
Studies investigating the neural bases of cognitive phenomena increasingly employ multialternative detection tasks that seek to measure the ability to detect a target stimulus or changes in some target feature (e.g., orientation or direction of motion) that could occur at one of many locations. In such tasks, it is essential to distinguish the behavioral and neural correlates of enhanced perceptual sensitivity from those of increased bias for a particular location or choice (choice bias). However, making such a distinction is not possible with established approaches.
We present a new signal detection model that decouples the behavioral effects of choice bias from those of perceptual sensitivity in multialternative (change) detection tasks. By formulating the perceptual decision in a multidimensional decision space, our model quantifies the respective contributions of bias and sensitivity to multialternative behavioral choices. With a combination of analytical and numerical approaches, we demonstrate an optimal, one-to-one mapping between model parameters and choice probabilities even for tasks involving arbitrarily large numbers of alternatives.
We validated the model with published data from two ternary choice experiments: a target-detection experiment and a length-discrimination experiment. The results of this validation provided novel insights into perceptual processes (sensory noise and competitive interactions) that can accurately and parsimoniously account for observers’ behavior in each task. The model will find important application in identifying and interpreting the effects of behavioral manipulations (e.g., cueing attention) or neural perturbations (e.g., stimulation or inactivation) in a variety of multialternative tasks of perception, attention, and decision-making.
Over the past century, major strides have been made in characterizing the phenomenology of attention in humans. As a result of this research, a variety of attention disorders, such as attention deficit disorder, autism and schizophrenia, can now be reliably diagnosed. However, the etiologies of these disorders remain poorly understood. Developing targeted therapies for treating such disorders requires a mechanistic understanding of how attention works at the level of cells and circuits. Here we review recent evidence for remarkable similarities in the phenomenology of spatial selective attention in birds and primates. These studies open up new avenues for research into the neural mechanisms that control attention. The brains of birds and primates share many neuroanatomical and functional features. Like primates, birds (especially chickens) are readily trained to perform behavioral tasks that yield precise, quantitative measures of decision-making. In contrast to primates, they are readily available and tractable for developing and applying cutting-edge experimental techniques. We expect, therefore, that research on avian species will greatly accelerate the discovery of neural mechanisms that underlie attention.
Voluntary control of attention promotes intelligent, adaptive behaviors by enabling the selective processing of information that is most relevant for making decisions. Despite extensive research on attention in primates, the capacity for selective attention in nonprimate species has never been quantified. Here we demonstrate selective attention in chickens by applying protocols that have been used to characterize visual spatial attention in primates. Chickens were trained to localize and report the vertical position of a target in the presence of task-relevant distracters. A spatial cue, the location of which varied across individual trials, indicated the horizontal, but not vertical, position of the upcoming target. Spatial cueing improved localization performance: accuracy (d′) increased and reaction times decreased in a space-specific manner. Distracters severely impaired perceptual performance, and this impairment was greatly reduced by spatial cueing. Signal detection analysis with an “indecision” model demonstrated that spatial cueing significantly increased choice certainty in localizing targets. By contrast, error-aversion certainty (certainty of not making an error) remained essentially constant across cueing protocols, target contrasts, and individuals. The results show that chickens shift spatial attention rapidly and dynamically, following principles of stimulus selection that closely parallel those documented in primates. The findings suggest that the mechanisms that control attention have been conserved through evolution, and establish chickens—a highly visual species that is easily trained and amenable to cutting-edge experimental technologies— as an attractive model for linking behavior to neural mechanisms of selective attention.
[8] Magnetic tracking of eye position in freely behaving chickens.
Research on the visual system of non-primates, such as birds and rodents, is increasing. Evidence that neural responses can differ dramatically between head-immobilized and freely behaving animals underlines the importance of studying visual processing in ethologically relevant contexts. In order to systematically study visual responses in freely behaving animals, an unobtrusive system for monitoring eye-in-orbit position in real time is essential. We describe a novel system for monitoring eye position that utilizes a head-mounted magnetic displacement sensor coupled with an eye-implanted magnet. This system is small, lightweight, and offers high temporal and spatial resolution in real time. We use the system to demonstrate the stability of the eye and the stereotypy of eye position during two different behavioral tasks in chickens. This approach offers a viable alternative to search coil and optical eye tracking techniques for high resolution tracking of eye-in-orbit position in behaving animals.
[7] Spatial probability dynamically modulates visual target detection in chickens.
The natural world contains a rich and ever-changing landscape of sensory information. To survive, an organism must be able to flexibly and rapidly locate the most relevant sources of information at any time. Humans and non-human primates exploit regularities in the spatial distribution of relevant stimuli (targets) to improve detection at locations of high target probability. Is the ability to flexibly modify behavior based on visual experience unique to primates? Chickens (Gallus domesticus) were trained on a multiple alternative Go/NoGo task to detect a small, briefly-flashed dot (target) in each of the quadrants of the visual field. When targets were presented with equal probability (25%) in each quadrant, chickens exhibited a distinct advantage for detecting targets at lower, relative to upper, hemifield locations. Increasing the probability of presentation in the upper hemifield locations (to 80%) dramatically improved detection performance at these locations to be on par with lower hemifield performance. Finally, detection performance in the upper hemifield changed on a rapid timescale, improving with successive target detections, and declining with successive detections at the diagonally opposite location in the lower hemifield. These data indicate the action of a process that in chickens, as in primates, flexibly and dynamically modulates detection performance based on the spatial probabilities of sensory stimuli as well as on recent performance history.
[6] Gamma oscillations are generated locally in an attention-related midbrain network.
Gamma-band (25–140 Hz) oscillations are a hallmark of sensory processing in the forebrain. The optic tectum (OT), a midbrain structure implicated in sensorimotor processing and attention, also exhibits gamma oscillations. However, the origin and mechanisms of these oscillations remain unknown. We discovered that in acute slices of the avian OT, persistent (~100 ms) epochs of large amplitude gamma oscillations can be evoked that closely resemble those recorded in vivo. We found that cholinergic, glutamatergic, and GABAergic mechanisms differentially regulate the structure of the oscillations at various timescales. These persistent oscillations originate in the multisensory layers of the OT and are broadcast to visual layers via the cholinergic nucleus Ipc, providing a potential mechanism for enhancing the processing of visual information within the OT. The finding that the midbrain contains an intrinsic gamma-generating circuit suggests that the OT could use its own oscillatory code to route signals to forebrain networks.
[5] Space coding by gamma oscillations in the barn owl optic tectum.
Gamma-band (25–140 Hz) oscillations of the local field potential (LFP) are evoked by sensory stimuli in the mammalian forebrain and may be strongly modulated in amplitude when animals attend to these stimuli. The optic tectum (OT) is a midbrain structure known to contribute to multimodal sensory processing, gaze control, and attention. We found that presentation of spatially localized stimuli, either visual or auditory, evoked robust gamma oscillations with distinctive properties in the superficial (visual) layers and in the deep (multimodal) layers of the owl’s OT. Across layers, gamma power was tuned sharply for stimulus location and represented space topographically. In the superficial layers, induced LFP power peaked strongly in the low-gamma band (25–90 Hz) and increased gradually with visual contrast across a wide range of contrasts. Spikes recorded in these layers included presumptive axonal (input) spikes that encoded stimulus properties nearly identically with gamma oscillations and were tightly phase locked with the oscillations, suggesting that they contribute to the LFP oscillations. In the deep layers, induced LFP power was distributed across the low and high (90–140 Hz) gamma-bands and tended to reach its maximum value at relatively low visual contrasts. In these layers, gamma power was more sharply tuned for stimulus location, on average, than were somatic spike rates, and somatic spikes synchronized with gamma oscillations. Such gamma synchronized discharges of deep-layer neurons could provide a highresolution temporal code for signaling the location of salient sensory stimuli.
[4] A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks.
Cognitively demanding tasks that evoke activation in the brain’s central-executive network (CEN) have been consistently shown to evoke decreased activation (deactivation) in the default-mode network (DMN). The neural mechanisms underlying this switch between activation and deactivation of large-scale brain networks remain completely unknown. Here, we use functional magnetic resonance imaging (fMRI) to investigate the mechanisms underlying switching of brain networks in three different experiments. We first examined this switching process in an auditory event segmentation task. We observed significant activation of the CEN and deactivation of the DMN, along with activation of a third network comprising the right fronto-insular cortex (rFIC) and anterior cingulate cortex (ACC), when participants perceived salient auditory event boundaries. Using chronometric techniques and Granger causality analysis, we show that the rFIC-ACC network, and the rFIC, in particular, plays a critical and causal role in switching between the CEN and the DMN. We replicated this causal connectivity pattern in two additional experiments: (i) a visual attention "oddball" task and (ii) a task-free resting state. These results indicate that the rFIC is likely to play a major role in switching between distinct brain networks across task paradigms and stimulus modalities. Our findings have important implications for a unified view of network mechanisms underlying both exogenous and endogenous cognitive control.
[3] An in-silico model of dynamic routing through neuronal coherence.
We describe a neurobiologically plausible model to implement dynamic routing using the concept of neuronal communication through neuronal coherence. The model has a three-tier architecture: a raw input tier, a routing control tier, and an invariant output tier. The correct mapping between input and output tiers is realized by an appropriate alignment of the phases of their respective background oscillations by the routing control units. We present an example architecture, implemented on a neuromorphic chip, that is able to achieve circular-shift invariance. A simple extension to our model can accomplish circular-shift dynamic routing with only O(N) connections, compared to O(N2) connections required by traditional models.
[2] Neural dynamics of event segmentation in music: Converging evidence for dissociable ventral and dorsal networks.
The real world presents our sensory systems with a continuous stream of undifferentiated information. Segmentation of this stream at event boundaries is necessary for object identification and feature extraction. Here, we investigate the neural dynamics of event segmentation in entire musical symphonies under natural listening conditions. We isolated time-dependent sequences of brain responses in a 10 s window surrounding transitions between movements of symphonic works. A strikingly right-lateralized network of brain regions showed peak response during the movement transitions when, paradoxically, there was no physical stimulus. Model-dependent and model-free analysis techniques provided converging evidence for activity in two distinct functional networks at the movement transition: a ventral fronto-temporal network associated with detecting salient events, followed in time by a dorsal fronto-parietal network associated with maintaining attention and updating working memory. Our study provides direct experimental evidence for dissociable and causally linked ventral and dorsal networks during event segmentation of ecologically valid auditory stimuli.
[1] The role of the basal ganglia in exploration in a neural model based on reinforcement learning.
We present a computational model of basal ganglia as a key player in exploratory behavior. The model describes exploration of a virtual rat in a simulated water pool experiment. The virtual rat is trained using a reward-based or reinforcement learning paradigm which requires units with stochastic behavior for exploration of the system’s state space. We model the Subthalamic Nucleus-Globus Pallidus externa (STN-GPe) segment of the basal ganglia as a pair of neuronal layers with oscillatory dynamics, exhibiting a variety of dynamic regimes such as chaos, traveling waves and clustering. Invoking the property of chaotic systems to explore state-space, we suggest that the complex exploratory dynamics of STN-GPe system in conjunction with dopamine-based reward signaling from the Substantia Nigra pars compacta (SNc) present the two key ingredients of a reinforcement learning system.
Miscellanea
Awards and recognitions
2014
Ramalingaswami Fellowship (transition award) for early career investigators, Department of Biotechnology, Govt of India.
2012
School of Medicine Dean’s Fellowship Award for Postdoctoral Research, Stanford University.
2010
Gatsby Cosyne Travel Fellowship (selected speaker), Computational and Systems Neuroscience (COSYNE) Meeting, Salt Lake City.
2007
NIH Human Brain Mapping Travel Award and Stanford BioX Travel Award, (selected speaker), 14th Annual Meeting of the Organization for Human Brain Mapping (OHBM), Chicago.
2004
Stanford Graduate (Smith) Fellowship. One of Stanford University's highest honors for incoming graduate students.
2004
President of India Gold Medal. Best academic record, graduating class of 2004, Indian Institute of Technology (IIT), Madras.
Ministry of Human Resources and Development (MHRD) Assistantship. Ranked fourth nationwide in the Graduate Aptitude Test of Engineering (GATE), Govt. of India.
2001-2004
Institute Merit Prize. Best academic record for four consecutive years in Aerospace Engineering, Indian Institute of Technology, (IIT) Madras.
1999
National Scholarships Scheme (NSS) Award. Top 0.1% nationwide score, All India Senior School Certificate Examinations, Govt. of India.
1997-2003
National Talent Search (NTS) Scholar, National Council for Educational Research and Training (NCERT), India.
1997, 1999
School topper in Secondary and Higher Secondary, Vidya Mandir (Adyar), Chennai, India.
Young Investigators' Meeting, Hyderabad India. Title:The role of gamma oscillations in sensory coding and selective attention: An avian midbrain model. One of four post-doctoral fellows selected to speak at the meeting.
2014
Center for Neural and Cognitive Sciences, Hyderabad, India (Title: ibid.)
Friday Seminar Series, Stanford Center for Mind, Brain and Computation and Department of Psychology Title: Modeling decisions among multiple alternatives: Decoupling bias from sensitivity effects in detection and attention tasks.
3rd Annual Stanford Postdoctoral Research Symposium: Empowering Future Leaders. Title:Visuospatial selective attention in chickens. One of ten post-doctoral fellow speakers selected from across Stanford University.
2013
Centre for Neuroscience (CNS), Indian Institute of Science (IISc), Bengaluru, India Title: Decoupling choice bias from perceptual sensitivity: A signal detection approach for multialternative detection and decision-making tasks.
2011
Society for Neuroscience (SfN, 41st annual meeting), Washington DC, USA Title:Towards a mechanistic understanding of the role of gamma oscillations in attention. Nano-symposium on "Functional Mechanisms of Attention by Animal"
2010
Computational and Systems Neuroscience (COSYNE), Salt Lake City, Utah, USA Title:Robust spatial working memory through inhibitory gamma synchrony. Ranked among the top 5% of over 400 submissions and selected for oral presentation.
2009
Bioengineering Forum, Stanford University Title:Gamma oscillations in the avian optic tectum.
2007
Organization for Human Brain Mapping (OHBM, 13th annual meeting), Chicago USA Title:A causal role for the right fronto-insular cortex in switching between executive-control and default-mode networks. Ranked among the top 5% of over 1500 submissions and selected for oral presentation.
2007
Neural Information Processing Systems (NIPS), Vancouver, Canada Title:An in-silico model of dynamic routing through neuronal coherence. Ranked among the top 10% of around 1000 submissions, and selected for oral spotlight presentation (delivered by co-first author, Brian Percival).
2007
National Center for Biological Sciences (NCBS), Bengaluru, India Title:fMRI acquisition and analysis: Techniques and tools.
2007
2nd Annual Stanford Symposium on Music, Rhythm and the Brain, Stanford University Title:The neuroscience of music perception explored through functional imaging and computational modeling.
Distinct networks in the forebrain are hypothesized to be involved in attention control: (i) a "dorsal" fronto-parietal network, comprising the dorsolateral prefrontal cortex and posterior parietal cortex; (ii) a "ventral" fronto-temporal network, comprising the ventrolateral prefrontal cortex and the temporo-parietal junction and (iii) a cingulo-opercular network, comprising the insula, anterior cingulate cortex and thalamus. As a Smith Fellow at Stanford, I investigated the dynamical interplay of these networks during task performance, and task-free resting state, with functional neuroimaging (fMRI) in humans.
Ventral and dorsal attention networks activate during auditory event boundaries
The real world presents our sensory systems with a continuous stream of undifferentiated information. Segmentation of this stream at event boundaries is necessary for identifying and attending to relevant objects that appear in the event stream. However, the neural processes mediating the segmentation of real-world stimuli remain unknown.
With subjects listening to auditory (musical) stimuli inside an fMRI scanner, we discovered that two distinct functional networks were active at event boundaries: a ventral fronto-temporal network associated with detecting salient events, followed in time by a dorsal fronto-parietal network associated with maintaining attention and updating working memory. This study provided direct experimental evidence for dissociable and functionally linked ventral and dorsal attention networks during event segmentation of ecologically valid auditory stimuli.
Sridharan D, Levitin DJ, Chafe CH, Berger J, Menon V. Neural dynamics of event segmentation in music: Converging evidence for dissociable ventral and dorsal networks.Neuron (2007) 55(3):521-32.
A role for the right fronto-insular cortex in cognitive control
Cognitively demanding tasks have been consistently shown to evoke switching between specialized networks in the brain: increased activation in the central-executive network (CEN) and decreased activation (deactivation) in the default-mode network (DMN). Yet, the neural mechanisms underlying this switch between activation and deactivation of large-scale brain networks are unknown.
Using chronometric techniques and Granger causality analysis, we showed that the right fronto-insular cortex (rFIC), a key node of the salience network, is likely to play a critical role in switching between distinct brain networks across various task paradigms and stimulus modalities. These findings could have important implications for a unified view of network mechanisms underlying both exogenous and endogenous cognitive control.
Sridharan D, Levitin DJ, Menon V. A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks.Proceedings of the National Academy of Sciences (2008) 105(34):12569-74.
Conducted at: Psychiatry and Behavioral Sciences, Stanford University
Years: 2004-07
Gamma oscillations in the midbrain attention network
Gamma-band (25-140 Hz) oscillations in the neuroelectric activity of the brain are reported extensively in the forebrain reiogns involved in attention control. It is increasingly recognized that a network of structures in the midbrain also contributes to attention control. My graduate research combined in vivo and in vitro electrophysiology to characterize network mechanisms of gamma oscillations in the avian midbrain.
Gamma oscillations in the optic tectum induced by visual and auditory stimuli in-vivo
Gamma-band (25–140 Hz) oscillations of the local field potential (LFP) are ubiquitous in the mammalian forebrain and hippocampus. The oscillations are evoked by sensory stimuli, and strongly modulated in amplitude during cognitive processes like attention. However, little is known about these oscillations in other, non-mammalian species or other brain structures.
The optic tectum (OT) is a key midbrain structure in a network of regions involved in gaze control, and attention. We discovered that presentation of spatially localized stimuli, either visual or auditory, evoked robust, space-specific, gamma LFP oscillations in the barn owl OT. In addition, spiking activity recorded in the OT showed robust phase-locking to the LFP in the low-gamma (25-90 Hz) band. Such gamma-synchronized discharges of OT neurons could provide a high-resolution temporal code for signaling the location of salient sensory stimuli.
Sridharan D, Boahen K, Knudsen EI. Space coding by gamma oscillations in the barn owl optic tectum. Journal of Neurophysiology (2011) 105: 2005-2017.
Gamma oscillations are locally generated in the midbrain network
In the previous study, we showed that gamma oscillations can be recorded from the midbrain network. Are these oscillations transmitted from forebrain areas, or are they locally generated within the midbrain? We sought to investigate the origins and mechanisms of these oscillations in-vitro.
We discovered that in acute slices of the avian OT, robust gamma oscillations could be evoked by electrical microstimulation, that closely resemble oscillations recorded in-vivo. The oscillations are generated and regulated by pharmacological mechanisms (cholinergic, GABAergic) and E-I circuits (including putative parvalbumin-positive inhibitory neurons), that are remarkably similar to those observed in the mammalian forebrain. The findings argue for a conserved, functional role for gamma oscillations over two hundred million years of evolution.
Goddard CA, Sridharan D, Huguenard JH, Knudsen EI. Gamma oscillations are generated locally in an attention-related midbrain network.Neuron (2012) 73(3):567-80 ( co-first authors).
Other studies
Sridharan D, Knudsen EI. Gamma oscillations in the midbrain spatial attention network: Linking circuits to functionCurrent Opinion in Neurobiology. Special Issue: Brian Rhythms and Dynamic Coordination (under review)
Conducted at: Department of Neurobiology, Stanford University
Years: 2008-11
Selective attention in birds
Attention control is mediated by distinct networks in the forebrain (the frontoparietal system) and the midbrain (the superior colliculus, and other midbrain nuclei). Among vertebrate species, the midbrain network is the most highly differentiated in birds, with cholinergic and GABAergic neurons organized into distinct nuclei and circuits. Birds are, therefore, a natural choice for identifying the contribution of these distinct circuits of the midbrain network to spatial attention.
Many studies of midbrain anatomy and physiology have provided indirect evidence for a critical role for these cholinergic and GABAergic circuits in spatial attention. Yet, the contribution of these circuits to attention control has never been directly tested in the behaving animal.
Chickens: A novel model for spatial vision and attention
For my post-doctoral work, I developed a highly-accessible, novel model for spatial vision and attention: Chickens!
Even newly hatched chicks are able to perform complex visuospatial behaviors, such as tracking a moving object on the screen (movie) and pecking on moving objects in a field of stationary objects. These behaviors are innate and do not require reinforcement, conditioning or prior exposure to the stimuli. These observations paved the way for developing a reinforcement learning protocol and apparatus that has enabled, for the first time, training chickens rapidly and reliably to perform complex spatial vision tasks (next).
Chickens flexibly bias behavior based on target spatial probability
The natural world contains a rich and ever-changing landscape of sensory information. To survive, an organism must be able to flexibly and rapidly locate the most relevant sources of information at any time. Humans and non-human primates exploit regularities in the spatial distribution of relevant stimuli (targets) to improve detection at locations of high target probability. Is the ability to modify behavior based on visual experience unique to primates?
We discovered that chickens, like primates, flexibly and dynamically bias their detection based on the spatial probabilities of target stimuli, as well as on recent performance history. This study was the first to demonstrate that chickens could be trained to perform complex spatial vision tasks with novel technology — a touch-screen based stimulus delivery and response acquisition system, along with real-time tracking of head position and orientation — developed specifically for measuring spatial vision behaviors in these birds.
Sridharan D, Ramamurthy DL, Knudsen EI. Spatial probability dynamically modulates visual target detection in chickens. Plos One (2013) 8(5): e64136.
Visuospatial selective attention in chickens
Top-down selective attention promotes intelligent, adaptive behaviors by enabling the selective processing of information that is most relevant for making decisions. Despite extensive research on attention in primates, it is unknown if non-primate species exhibit the capacity for selective attention. Leveraging techniques developed in the previous study, we asked if chickens would exhibit the behavioral hallmarks of selective attention.
We discovered that chickens shift spatial attention rapidly and dynamically, following principles of stimulus selection that closely parallel those documented in primates. Birds exhibited dramatic improvements in performance (percent correct and sensitivity, d') when provided with a top-down spatial cue in a visuospatial target localization task. The findings suggest that the mechanisms that control attention have been conserved through evolution. This study established chickens — a highly visual species that is easily trained and amenable to high-resolution experimental technologies — as an attractive model for elucidating the neural mechanisms of selective attention.
Sridharan D, Ramamurthy DL, Schwarz JS, Knudsen EI. Visuospatial selective attention in chickens.Proceedings of the National Academy of Sciences (2014) 111(19): E2056-2065.
The midbrain's role in spatial attention control (ongoing work)
The vertebrate midbrain contains specialized circuits, comprising cholinergic and GABAergic nuclei, that are thought to be critically involved in spatial attention control. However, the contribution of these circuits has never been tested in the behaving animal.
We have developed a novel, cued orientation discrimination task for chickens (movie): Birds are trained to give a Go (peck toward) response when presented with a horizontal grating, and a NoGo (peck center) response with a vertical grating. We will test how inactivating key components of the bird's midbrain network affects the birds' behavior in this perceptual discrimination task. The results will establish the contribution of the midbrain to spatial attention.
Sridharan D, Schwarz JS, Knudsen EI. Selective attention in birds.Current Biology (2014) 24(11): R510-513 (Primer).
Other studies
Schwarz JS, Sridharan D, Knudsen EI. Magnetic tracking of eye position in freely behaving chickens.Frontiers in Systems Neuroscience (2013) 7(91):1-8.
Conducted at: Department of Neurobiology, Stanford University
Years: 2010-14
Neuromorphic computational models of oscillations and synchrony
Oscillations in the neuroelectric activity of the brain are widely observed in various brain structures during sensory and cognitive processes in diverse species, from insects to mammals. Yet, it is unknown if these oscillations are merely epiphenomena, or play a causal role in neural processing. My graduate research explored a variety of computational models, implemented in hardware and software, that demonstrated key putative functional roles of these oscillations in neural information processing.
Inhibitory gamma synchrony promotes robust working memory
Persistent firing and gamma (25-140 Hz) oscillations are ubiquitous neural correlates of attention and working memory. In this work, we developed a model in which these gamma oscillations have a direct role in the robust maintenance of persistent information during working memory.
We propose a simple mechanism operating at the timescale of milliseconds to seconds that can reversibly create functional homogeneity in the firing landscape: inhibitory gamma synchrony. Upon imposing an external (background) inhibitory rhythm in the gamma range, model pyramidal neurons tend to synchronize to the rhythm, and equalize their firing rates. The resulting firing homogeneity permits persistent, localized firing with little drift, and could facilitate robust maintenance of information in working memory. More information here
We tested this theory using a neuromorphic VLSI chip (Arthur and Boahen, 2006) with 1024 excitatory silicon neurons arranged in a 32x32 grid with 256 interleaved inhibitory neurons. The simulation, shown alongside, shows the spatial profile, and spike rasters of the silicon neurons recorded in real time from the chip. A clear network rhythm in the gamma-band is apparent in the peristimulus time histogram (lower trace).
Additionally, we developed a custom interface for interacting with the chip in real-time, thanks two amazing Python projects: i) the Cython/Pyrex compiler, and ii) the Pyglet multimedia library. More information about this soon!
Sridharan D, Millner S, Arthur J, Boahen K. Robust spatial working memory through inhibitory gamma synchrony. Frontiers in Neuroscience Conference Proceedings (2010) doi: 10.3389/conf.fnins.2010.03.00012.
Selected for a contributed talk at the Computational and Systems Neuroscience (COSYNE) Meeting, Salt Lake City, 2010.
Neuronal communication through coherence enables dynamic routing of information
The ability to flexibly switch attention among various relevant stimuli in a dynamically changing environment is fundamental to adaptive survival. Information about the most relevant stimulus at each moment in time must be dynamically routed to higher processing and decision regions, in order to enable these prioritized stimuli to be processed more efficiently. However, the mechanisms by which the brain accomplishes dynamic routing of information remain unknown.
We developed a neurobiologically plausible model to implement dynamic routing using the concept of neuronal communication through coherence. The model had a three-tier architecture: a raw input tier, a routing control tier, and an invariant output tier. The correct mapping between input and output tiers is realized by an appropriate alignment of the phases of their respective background oscillations by the routing control units. We implemented this model on a neuromorphic silicon chip, and demonstrate that the model is able to achieve circular-shift invariance. The model, and neuromorphic architecture, demonstrate a key role for oscillations in flexibly remapping neural information according to environmental context or stimulus priority during attention.
Sridharan D, Percival B, Arthur J, Boahen K. An in-silico model of dynamic routing through neuronal coherence.Advances in Neural Information Processing Systems (NIPS) (2008) 20:1401-1408.
Chaotic dynamics drive exploration in a neural model of the basal ganglia
Fundamental to reinforcement learning is the ability to effectively explore novel environments. Yet, the neural circuits and mechanisms by which animals are able to efficiently navigate or forage in complex environments remain unknown.
We present a computational model of the basal ganglia as a key player in exploratory behavior. The model describes exploration of an animat (a virtual animal) in a simulated water pool experiment. The animat is trained using a reward-based or reinforcement learning paradigm which requires units with stochastic behavior for exploration of the system’s state space. We model the Subthalamic Nucleus-Globus Pallidus externa (STN-GPe) segment of the basal ganglia as a pair of neuronal layers with oscillatory dynamics, exhibiting a variety of dynamic regimes such as chaos, traveling waves and clustering. Invoking the property of chaotic systems to explore state-space, we suggest that the complex exploratory dynamics of STN-GPe system in conjunction with dopamine-based reward signaling from the Substantia Nigra pars compacta (SNc) present the two essential ingredients of a reinforcement learning system.
Sridharan D, Prashanth PS, Chakravarthy VS. The role of the basal ganglia in exploration in a neural model based on reinforcement learning. International Journal of Neural Systems (2006) 16(2):111-24.
Other studies
Sridharan D, Knudsen EI. Selective disinhibition: A robust mechanism for effecting target priority by selective attention. Vision Research. Special Issue: Computational Models of Visual Attention (under revision).
Conducted at: Department of Bioengineering, Stanford University
Years: 2007-10
Theoretical frameworks for attention behaviors
Rigorous behavioral protocols and psychophysical theories are essential to understand how cognitive phenomena, such as selective attention, emerge from the action of neurons and neural networks. My most recent work involves developing normative theoretical frameworks that enable establishing a mechanistic link between neurons and behavior.
Distinguishing bias from sensitivity effects in multialternative tasks
Studies investigating the neural bases of cognitive phenomena, such as attention, increasingly employ multialternative detection tasks. These tasks seek to compare the subject's ability to detect a target stimulus among multiple, distinct locations (e.g., at a location cued for attention vs. other locations). In such tasks, it is essential to distinguish the behavioral and neural correlates of enhanced perceptual sensitivity (increased signal to noise) from those of enhanced bias (altered decision criteria) for a particular location or choice. However, making such a distinction is not possible with established approaches.
We developed a new signal detection model, the m-ADC model that decouples the behavioral effects of criterion (bias) changes from those of sensitivity changes in multialternative (change) detection tasks. By formulating the perceptual decision in a multidimensional decision space, our model quantifies the respective contributions of bias and sensitivity to multialternative behavioral choices. With a combination of analytical and numerical approaches, we have shown an optimal, one-to-one mapping between model parameters and choice probabilities even for tasks involving arbitrarily large numbers of alternatives. We have also validated the model by demonstrating its ability to accurately fit human behavior in two ternary choice tasks: a target detection task and a length discrimination task.
The m-ADC model will find important application in identifying and interpreting the effects of behavioral manipulations (e.g., cueing attention) or neural perturbations (e.g., stimulation or inactivation) in multialternative tasks of perception, attention, and decision-making.
Our model provides an essential tool for identifying the neural correlates of bias and sensitivity changes in attention tasks. More information about the model, and Matlab code for behavioral analysis is available here
Sridharan D, Steinmetz NA, Moore T, Knudsen EI. Distinguishing bias from sensitivity effects in multialternative detection tasks. Journal of Vision (2014) 14(9):16, 1–32.
Applying behavioral frameworks to reveal mechanisms of brain function (ongoing work)
I am currently working on extending the m-ADC model and applying it to understand neural phenomena. First, the m-ADC analytical framework must be revised for modeling optimal decision-making in attention tasks, in which the perceptual sensitivity (d') at the attended location can be significantly different from that at other, unattended locations. Second, we will apply the m-ADC model to clarify the mechanistic role of the superior colliculus, a central structure in the midbrain network, to selective attention. Finally, we will apply the model to understand the mechanistic basis of firing rate modulation observed in a forebrain visual area (V4) during attention tasks. Information about each of these studies will be posted here as they are published.
Sridharan D, Steinmetz NA, Moore T, Knudsen EI. A unified framework for multiple-alternative detection in birds and primates. Journal of Vision: Vision Sciences Society (VSS) Conference Proceedings. (2013) 13(9):629. doi: 10.1167/13.9.629
Sridharan D, Steinmetz NA, Moore T, Knudsen EI. Modeling decisions among multiple alternatives: A normative framework for detection and attention tasks. (in preparation).
Sridharan D, Steinmetz NA, Moore T, Knudsen EI. The superior colliculus controls spatial choice bias during visuospatial selective attention. (in submission).
Conducted at: Department of Neurobiology, Stanford University
![[Icon]](./img/modals/jov_icon_large.gif){kind=link}
![[Cover]](./img/modals/mod2_jnphyscover.png){kind=link}
![[Cover]](./img/modals/neuroncover.gif){kind=link}