The ability to flexibly and adaptively integrate information from a variety of sources is a fundamental feature of brain function, from higher cognition to sensory and motor processing. Even a simple behavior such as reaching to a target relies on the integration of multimodal sensory signals and, moreover, exhibits rapid adaptation in response to changes in these signals. Our research uses reaching and similar goal-directed movements as a model system for understanding these abilities and their underlying neural mechanisms.

The lab employs a combination of complementary approaches:


The ultimate goal of our work is to discover cortical mechanisms for sensorimotor integration and learning. Physiological studies aimed at this goal are the current primary experimental focus of the lab. Three general approaches are being used:

  • Large-scale cortical recordings: We can simultaneously record from large neural populations using multiple 96-channel electrode arrays. Array recordings allow us to perform quantitative analyses at the level of the population responses in multiple brain areas. For example, see Dadarlat, O'Doherty, & Sabes (2015).
  • Manipulating cortical activity patterns: A major thrust of the lab is the development of techniques to both measure changes in network dynamics and drive those changes by directly manipulating patterns of cortical activity. Two approaches are being developed in parallel: patterned electrical stimulation and patterned light stimulation in tissue expressing light-sensitive ion channels ("optogenetics"). For example, see Yazdan-Shahmorad et al. (2016)
  • Human physiology: We have access to a variety of human neurophysiological tools, and occasionally make use of them. For example, read here about our fMRI study of adaptive Bayesian priors for reaching.


We use computational and theoretical models to link our understanding of brain and behavior. Two levels of modeling are used:

  • Models of behavior: Predictive models, typically cast in statistical or control-theoretical terms, provide intuition about why the behavior is the way it is. For example, see our work on the statistics of motor variability in Chaisanguanthum, Shen, and Sabes (2014).
  • Networks models: Often, a simple network model can be found that yields an approximation to our behavioral models. These network models provide intuition and about the circuit-level mechanisms that underlie behavior. Furthermore, these models generate testable hypotheses about the dynamics of cortical networks, and we use these models to design our physiological experiments. For example, see Makin, Dichter, and Sabes (2015).


With human psychophysics (or quantitative behavioral studies), we identify behavioral phenomena that illustrate important features of sensorimotor processing. Our goal is to find phenomena that are experimentally tractable for human and animals and are amenable to theoretical/computational modeling. For example, see our work on sensory integration for reaching planning and visuomotor adaptation, Verstynen and Sabes (2011).