Neuroculture: On the implications of brain science for understanding
Edmund T. Rolls, Oxford Centre for Computational Neuroscience
Oxford University Press (2012)
pp. 1-27, single neurons and neuron populations
Summary and review of the above secttion
Keywords: CA3, single neurons, neuron populations, neuron selectiveness
Rolls starts his book by arguing that there are too few genes to specify the total connectivity of the brain. Genes can only specify general rules such as the types of neuron in particular brain regions having some connection to one another. An important area such as CA3 in the hippocampus is suggested to require only 20-50 genes. The full development of the brain beyond these general specifications relies on inputs and learning from the environment and experience. Within this system gene-defined ‘reinforcers’ are seen as having influence on behaviour.
Studies have shown that the firing rate of neurons codes for information about stimuli. For instance, neurons in the primary visual cortex have firing rates related to the orientation of an edge, while neurons in the primary taste cortex have firing rates related to the concentration of taste. In the case of taste, the firing rate also correlates to the subjectively rated level of intensity.
Going beyond the primary cortex, sensory neurons can do more than just encode edges or intensities. The secondary taste cortex in the orbitofrontal cortex conveys information about what a particular taste is. A neuron may increase its firing rate in response to, for instance, glucose, but not for other stimuli such as salt.
The selectiveness of the neuron may be quite narrow, so that the neuron that codes for glucose may show only a few spikes for fruit juice. In the case of ‘face neurons’ located in the inferior temporal cortex, the orbitofrontal and the amygdala, individual neurons respond strongly to a few faces, at an intermediate level to a few other faces, and hardly at all to many stimuli.
In the taste cortex, the firing rate of neurons can decrease as the requirement for the food becomes satiated, indicating that the firing rate does not encode an actual description of the food. The subjectively rated pleasantness of the food correlates to the firing of the neurons, which both diminish as more of the food is eaten. The firing encodes both the brain reward value and the subjective pleasantness of the stimuli.
The author makes an important distinction as regards how neurons code for stimul. The firing rate of the neuron determines how much it contributes to a particular representation that is transmitted by a ppulation of neurons. The firing rate of neurons codes for the sterength of the stimuli, but which of the neurons are firing codes for which stimulus is present.
Spatial view neurons: These are seen to have an important role in the hippocampus. There are neurons in this system, which increase their firing rate when a particular part of the spatial environment is being viewed. The same firing response occurs for different angles of viewing indicating that the neurons represent coordinates out there in space (allocentric or world coordinates). This is suggested to be part of a system for remembering where objects have been seen. Overall the firing rate of single neurons are important for knowing how objects in the world are separated.
In a sparse representation only a small proportion of neurons are involved at any one time, in one study of neurons in the inferior temporal cortex, only a few stimuli produced high firing rates in particular neurons. This sparse distribution is seen as typical of the higher visual cortex, the taste and flavour related areas in the insula and the orbitofrontal and the spatial neurons in the hippocampus. This does not mean the neurons don’t respond at all to other stimuli, but they do respond much more to some stimuli than to others.
A stimulus is encoded by the firing rate distribution of a population of neurons. The code is seen as being read by the receiving neurons as the product of firing rates and synaptic weights; the slightly different weightings are the result of how previous inputs are stored in terms of the synaptic weightings. A neuron receives typically around 10,000 synaptic inputs from many different neurons. Each input at each synapse is weighted for the strength of the synapse, and the total input is the product of the thousands of differently weighted synapses. Each synapse on the neuron produces a small effect depending on the strength of the synapse. The depolarisation of the neuron is described as the sum of the number of inputs and the synaptic weight, as the dot product of these two vectors. If the input is not matched to the strengthened synapses there will be a weak response. The all-or-nothing aspect of neuron firing has the benefit of excluding weaker inputs.