Donald Doherty's blog

Category: Brain Science

  • Is There a Model of the Environment in the Brain?

    Typical approaches for studying neural coding focus on information transmission in neural circuits by quantifying how easily a stimulus can be recovered from evoked neural responses. These studies provide statistical descriptions of the mapping between stimuli and evoked responses.

    An alternative approach assesses the “statistical optimality of an internal model for probabilistic inference.” As best I can tell this approach considers a sensory system to hold a statistically optimal internal model of the environment if the neuronal activity evoked by a stimulus closely matches spontaneous activity in the system. In a new paper that takes this approach (“Spontaneous Cortical Activity Reveals Hallmarks of an Optimal Internal Model of the Environment” published January 7, 2011 in Science) the authors argue that neural activity is the result of the interaction between an internal model of the environment, embedded in underlying neural circuits, and the sensory input.

    Note: The paper under review has an associated 23 page PDF file containing supplementary material. Download the supplementary material from here.

    In this paper, they describe experiments in the primary visual cortex that test if evoked neuronal responses to natural visual stimuli closely match spontaneous neuronal responses. The analysis is formally based on the concepts of the likelihood of features and the prior distribution of features. The likelihood of features describes the probability with which any given input image can be expected to arise from a particular combination of features. The prior distribution of features is the probability with which any particular combination of features can be expected to occur.

    Spontaneous activity in the sensory cortex reflects the prior distribution of features. In contrast, stimulus evoked activity in the sensory cortex reflects the posterior distribution, which describes the probability that any given combination of features may have given rise to a particular input. The posterior distribution can be computed by Bayes’ rule from the likelihood of features and the prior distribution of features.

    The authors carried out the tests to see if evoked neuronal responses to natural visual stimuli closely matched spontaneous neuronal responses in animals at four different ages:

    • After the eyes first open.
    • After the maturation of orientation tuning and long range horizontal connections in the primary visual cortex.
    • In the young adult with a fully matured primary visual cortex.
    • In the older adult with a fully matured primary visual cortex.

    They found that the similarity between the spontaneous and evoked neuronal activities in the primary visual cortex increased with age and was specific to responses evoked by natural scenes. Their interpretation was that this showed a progressive “adaptation of internal models to the statistics of natural stimuli at the neural level.”

  • Electrical Synapses are Important for Learning and Memory

    Some neurons have direct electrical connections with each other through what are known as gap junctions. Gap junctions are composed of proteins that cross the cell membranes of adjoining cells and enable current to flow between them. Recent studies have shown that gap junctions can help enable oscillatory activity amongst neurons but otherwise their contribution to specific processes in the brain remains unclear. The recent paper “Electrical Synapses Control Hippocampal Contributions to Fear Learning and Memory” (published January 7, 2011 in Science) reports that gap junctions can be an important factor in learning and memory.

    Gap junctions are found between inhibitory (GABAergic) interneurons in the dorsal hippocampus and medial septum. These neurons drive hippocampal theta rhythms. Blocking gap junctions in the dorsal hippocampus disrupts theta rhythms and learning. The authors hypothesize that, since place cells fire in relation to the theta cycle, disrupting theta may abolish the temporal codes for locations. The role of gap junctions in signal processing has generally taken a back seat to chemical synapses. It’s exciting to see more information on the importance of gap junctions beginning to emerge.

    Other related blog posts:

    Appearance of High Frequency Oscillations When Rewarded

    Sensation and Location in the Hippocampal Formation

  • An Assessment of a Rate Coding Paper

    Okay, I’ve spent more time with the paper (and supplementary material) I first mentioned in my blog post three days ago (blog post “Coding in the Brain, Paper Bloat, and the Need to Change the Way Papers are Published“). The authors’ conclusion was that the “cortex is likely to use primarily a rate code.”

    Note: The paper is titled “Sensitivity to perturbations in vivo implies high noise and suggests rate coding in cortex” and was published July 1, 2010 in Nature. Be sure to also look at the 42 pages of supplementary material available here!

    On my initial reading of the paper I thought I must have missed the new empirical and/or theoretical evidence that led to their conclusion. As it happens there is none. Their conclusion, which they hedge at several points, is based on a circular argument.

    A perturbation was elicited by the researchers, consisting of a single extra spike in one neuron, that produced approximately 28 additional spikes in its postsynaptic targets. The spike produced a detectable increase in firing rate in the local network. The observed amplification was characterized by intrinsic variations in membrane potential on the order of 2.2 to 4.5 millivolts.

    The authors concluded that since the additional spike resulted in stimulus independent variations in membrane potential, the variations in membrane potential “are pure noise, and so carry no information at all.” First note that the variations in membrane potential are stimulus independent because the perturbation – the spike elicited by the research team – is stimulus independent.

    So really the argument comes down to the following:

    • a) Neurons and neuronal circuits are exquisitely sensitive to signals.
    • b) The brain, or at least the neocortex, is a noisy system.
    • Therefore c) the brain, or neocortex, must be using rate coding.

    This adds nothing to the question of what kind or kinds of coding are being used by the cerebral cortex. The same data could be used to make the following argument:

    • a) Neurons and neuronal circuits are exquisitely sensitive to signals.
    • b) The brain, or at least the neocortex, responds in a spatially and temporally precise manor to a single presynaptic spike.
    • Therefore c) the brain, or neocortex, must be using temporal coding.

    I’ll conclude with just one more point about the 42 pages of supplementary material and the fact that they did not publish their computational models or data to open online repositories.

    The team investigated the relationship between postsynaptic currents and the probability of eliciting an extra spike by constructing a biophysically realistic model of a layer 5 pyramidal cell based on the existing model Spike Initiation in Neocortical Pyramidal Neurons (Mainen et al 1995) published in the SenseLab ModelDB repository. (The model and its associated paper are briefly reviewed in my blog post “Standard Neocortical Pyramidal Neuron Model.”) The authors made three changes to the existing model:

    • The temperature was increased from 23 degrees centigrade to 37 degrees centigrade.
    • All membrane conductance values were multiplied by 3.
    • The passive reversal potential was set to -75 millivolts rather than -70 millivolts.

    These changes are easy enough for the reader to make in the original model by Mainen and colleagues. However, the authors specified in the supplementary material that the way they mimicked neuron activity as it would be while the neuron was part of the brain in a living animal “by bombarding the model neuron with synaptic input sufficiently large to generate voltage fluctuations on the order of 3 millivolts at the soma and a firing rate of 2.6 Hertz.” How, exactly, did they set this up so that the reader may replicate their results?

    Other related blog posts:

    Coding in the Brain, Paper Bloat, and the Need to Change the Way Papers are Published

    Standard Neocortical Pyramidal Neuron Model