I. Neural coding of posture & behavior
A major goal of the lab is to understand how the brain represents pose and movement of the body through space, and how elementary motions are synthesized into meaningful behavior. To do this we utilize 3D tracking of freely moving rodents while recording from groups of neurons during natural behavior. We study rodents because they are agile and are amenable to large-scale neural recordings. We work hard to ensure they are healthy and happy for our experiments, which produces higher quality behavioral data. This approach has so far produced new insights into how the cortex, the outer-most structure of the brain, encodes 3D posture of the head and body during unrestrained movement.
Here is a short explanation of how we combined 3D tracking with neural recordings to quantify postural tuning in the posterior parietal cortex (PPC) and frontal motor cortex (M2), areas critical for coordinating bodily movements through space. This work was just published in November 2018. A more complete, illustrated telling of this story by PhD student Bartul Mimica can be found here.
Figure by Tuce Tombaz / Kavli Institute for Systems Neuroscience.
(1) Mark the back and head of the rat using retro-reflective markers.
(2) Three-dimension tracking reveals a world of behavioral data missed by 2D approaches. Left: 2D tracking of the head (the traditional approach). Right: same recording, with 3D tracking of the head and back, gives a far richer sense of the animal's behavior (video by Bartul Mimica / Kavli Institute for Systems Neuroscience).
(3) We can record neural spiking in freely-running rats while recording behavior with a 6-camera near infrared tracking system (Optitrack, Inc.)
(4) The tuning properties of the cells can then be visualized during natural behavior. So far, we have found that posterior parietal and frontal motor cortices are dominated by 3D postural signals, exemplified by the cells below:
We are using a similar approach to study 3D behavioral tuning in several regions of cortex in freely behaving mice using miniature head-mounted microscopes. These allow us to record from 100's of individual neurons at a time, only instead of recoding individual spikes, the miniscopes record neural activity as fluorescent flashes via a technique known as calcium imaging.
3D models for tracking the head and back in mice
A prior attempt at whole-body 3D tracking, which we later simplified to the head and back
In vivo calcium imaging in a freely-behaving mouse
II. Social learning, encoding the posture of others
Another major point of interest in the lab is social cognition, action understanding and observational learning. Observational learning is advantageous in the animal kingdom because it allows an individual to learn whether something is safe or not by watching a cohort and avoiding any risk to themself. This type of learning has been reported in species ranging from chimpanzees to pigeons (here is one of many reviews), yet the network mechanisms that make it possible are only beginning to be understood. Far more of the underlying neuroscience is known, for example, for how your brain allows you to see or know where you are located in space.
To get at where and how the brain learns new behaviors from conspecifics, we are using behavioral paradigms coupled with brain stimulation, pharmacology, genetic and anatomical approaches. One eventual goal, for example, would be to see how one animal's brain encodes the 3D posture of another. More to come on this...
Conceptual design for an observational learning paradigm in which one animal learns new patterns of behavior by watching a cohort.