This site is no longer being updated.

My primary scientific interest lies in understanding how the brain forms percepts and uses them to make decisions, especially in the visual domain. In particular, I am interested in how the brain's perceptual beliefs about the outside world are represented by the responses of populations of cortical neurons and how their spiking activity gives rise to percepts and decisions. To that end I construct mathematical models that aim to explain neural responses and behavior.
Key concepts in the context of my work are perceptual decision-making, probabilistic inference, neural sampling, noise correlations, choice probabilities, population responses, optimal linear read-out, feedforward, recurrent and top-down processing, covert attention, psychophysical kernel.

Probabilistic inference and neural sampling
In order to draw conclusions, or inferences, about the outside world, the brain has to combine sensory information with its learnt knowledge about the structure of the external world. How this is implemented in the brain is still unknown. By generating predictions for classic perceptual tasks, I test the hypothesis that the brain performs probabilistic inference by sampling, i.e. that neuronal activity can be interpreted as samples from a generative model of the world that the brain has previously learnt.

Population (de)coding and perceptual decision-making
How many sensory neurons contribute to a particular decision, how are they being read out (e.g. optimal or not) and which neurons are they? We have made significant progress recently towards answering these questions by deriving the analytical relationship between noise correlations, choice probabilities and read-out weights. This will allow us to answer two of these questions as soon as multi-electrode recordings from behaving animals become available, i.e. very soon.
Applying this framework to neural recordings from MT during a dual motion direction and binocular discrimination task while area V2 was being cooled, we could constrain the origin of the noise correlations in MT.

Binocular vision
Depth perception from binocular images is an exemplary model system for studying how the brain extracts information not explicitly present in its (2D) inputs. I have been particularly interested in understanding what feedforward computations might underlie the observed neurophysiology and how much information different binocular neuron types contain about depth.

Natural image statistics
Understanding the statistics of the natural world is important for understanding the properties of early sensory processing. Traditionally, this argument has been made in the context of efficient coding (Barlow) but what learning principle (objective function) is responsible for the properties of early sensory neurons, e.g. their receptive fields in the case of visual neurons, is still an open question and active field of research. Ultimately, this question is related to what generative model the brain has learnt for its sensory inputs.

This site is no longer being updated.

Poster at AREADNE: "Top-down attention in a probabilistic inference framework" in collaboration with Pietro Berkes and Josef Fiser: http://areadne.org/

Paper out in eLife: "A neural basis for the spatial suppression of visual motion perception" in collaboration with Dave Liu and Chris Pack: https://elifesciences.org/content/5/e16167

Paper out in Neuron: "The implications of perception as probabilistic inference for correlated neural variability during behavior" in collaboration with Pietro Berkes and Jozsef Fiser: author link

Spring School:
"The role of simulations in neuroscience". Co-organized and co-taught a week-long seminar on the philosophy & science of simulations together with Philipp Berens & Eckhart Arnold, including visits to HBP co-director Felix Schürmann and chief critic Alex Pouget.

Paper accepted at Neuron: "The implications of perception as probabilistic inference for correlated neural variability during behavior" in collaboration with Pietro Berkes and Jozsef Fiser: http://arxiv.org/abs/1409.0257

Cosyne 2016
Two posters accepted
Invited Talk in workshop on "Form and function of choice-related feedback signals in decision making"

New paper: "A modality-specific feedforward component of choice-related activity in MT" in collaboration with Alexandra Smolyanskaya, Stephen G. Lomber and Richard T. Born (Neuron)

Cosyne 2015
Poster on "Choice probabilities, detect probabilities, and read-out with multiple neuronal input populations"
Workshop Talk on "The source of sensory & decision-related activity in area MT"

SfN 2014
Poster on "On the relationship between stimulus-evoked and choice-related responses and correlations during perceptual decision-making in a probabilistic inference framework"

In Sept 2014 moved to Rochester and started my own group as an assistant professor. I'm looking to recruit postdocs, graduate students and undergraduates, so please be in touch if you'd like to work with me. 

Bernstein Conference 2014
Talk titled "Inferring the brain's task-strategy from neuronal responses during perceptual decision-making"
Gordon Research Seminar on "Neurobiology of Cognition" 2014: Talk
Gordon Research Conference on "Neurobiology of Cognition" 2014Poster

Sloan-Swartz Meeting 2014
Talk titled "Perceptual decision-making as probabilistic inference: implications for the correlations between stimulus, neuronal responses, and behavior."

2014, March 12, 12pm: Neurolunch, Center for Brain Sciences, Harvard UniversityTalk on Top-down correlations in a probabilistic inference framework

2014, March 10, 11am: Computational Systems Club, Brandeis UniversityTalk on Top-down correlations in a probabilistic inference framework

Cosyne 2014
Poster at main meeting on "Good noise or bad noise: The role of correlated variability in a probabilistic inference framework": Poster PDF

2014, Feb 14, 9:30am: Systems Club, Harvard Medical School: Talk on Top-down correlations in a probabilistic inference framework

2014, Jan 21, 9:15am: BCS, Rochester University, NY: Talk on Choice probabilities and neural sampling
New publication 
We have compared the implications of slowness and sparseness objective functions for complex cell learning in V1. In contrast to earlier reports, we found that their effects for the shape of the receptive fields is very different.

2013, Dec 6, 3pm: Center for Neural Science, New York University, NYC: Talk on Choice probabilities and neural sampling

SfN 2013 

2013, Oct 25Center for Theoretical Neuroscience, Columbia University, NYC: Talk on Neural sampling

Bernstein 2013 Workshop: "Probabilistic inference with sensory neurons"
As part of this year's Bernstein Conference, Jozsef Fiser and I will organize a half-day workshop on probabilistic inference in the sensory system.

2013, Aug 29: Systems Neuroscience Brown Bag, Washington University, St. Louis:  Talk on Choice probabilities and neural sampling

2013, Jun 13: Bernstein Center for Computational Neuroscience, Berlin: Talk on Choice probabilities and neural sampling

Barcelona 2013 Workshop: "Noise in decision-making"
I gave a talk on neural sampling in the context of perceptual decision-making. Interestingly, Bruce Cumming showed unpublished data from multi-electrode recordings in V1 while a monkey was performing a 2AFC task that appears to confirm a central prediction of our work. Matthias Bethge presented our published work on noise correlations, read-out weights and choice probabilities.

2013, Apr 26: Systems Club, Department of Neurobiology, Harvard University: Talk on Neural sampling

Cosyne 2013
Presented two posters on neural sampling and perceptual decision-making

F1000 Recommendations
Brent Doiron, Bruce Cumming and Dora Angelaki have recommended our recent paper on noise correlations, read-out weights and choice probabilities.

New publication
We're the first to derive the analytical relationship between noise correlations, read-out weights and choice probabilities for a linear read-out. We show how this relationship can be used to infer the read-out weights in a given task from multi-electrode recordings, and how the optimality of the read-out can be tested from single-electrode recordings.