Dendrites and computation
The thousands of inputs a single cell receives can interact in complex ways that depend on their spatial arrangement and on the biophysical properties of their respective dendrites. For example, operations such as coincidence detection, pattern recognition, input comparison, and simple logical functions can be carried out locally within and across individual branches of a dendritic tree. Harnett addresses the hypothesis that the brain’s computational power arises from these fundamental integrative operations within dendrites. He focuses in particular on sensory processing and spatial navigation, with the goal of understanding the mechanistic basis of these brain functions.
Two-photon fluorescent image of of a single pyramidal neuron in a living brain slice from the rodent subiculum during simultaneous patch-clamp recording of somatic and dendritic electrical activity. Image: Mark Harnett
If integrative operations within neurons represent the building blocks for computations, then plasticity in the biophysical properties of individual neurons could provide a potent means for either reinforcing or changing neural processing algorithms. Most current models for how the brain learns are based on the concept of spike-timing-dependent plasticity, in which the relative timing of action potentials in presynaptic and postsynaptic neurons causes synapses to become either stronger or weaker. The complexity of dendritic processing, however, suggests many other possible mechanisms by which the function of neural circuits could be altered by experience. Harnett plans to explore this possibility using electrical and optical recording in behaving rodents and in vitro preparations to understand how changes in cellular properties lead to altered computations and thus to modification of behavior through learning.
Cognitive disorders such as autism and intellectual disability are often characterized by changes in the number, distribution, and shape of dendritic spines, the tiny bud-like protrusions where the majority of excitatory synapses are located. Harnett’s previous work suggests that changes in spines are likely to have important functional consequences for neural information processing and hence for computations performed in the affected circuits. At MIT he plans to study this directly using mouse genetic models of human brain disorders. By investigating how anatomical abnormalities alter dendritic operations, he hopes to generate biophysical targets for experimental manipulation. The eventual goal is to causally link structural and functional changes at the cellular level with the aberrant computations that lead to pathological behaviors.
Mark Harnett joined the McGovern Institute in 2015 and is currently the Fred and Carole Middleton Career Development Professor in the Department of Brain and Cognitive Sciences. Mark received his B.A. in Biology from Reed College in Portland, Oregon and his Ph.D. in Neuroscience from the University of Texas at Austin. Prior to joining MIT, he was a postdoctoral researcher at the Howard Hughes Medical Institute’s Janelia Research Campus in Ashburn VA where he worked with Jeff Magee.