Abstract

The nervous system performs computations to process information that is biologically important. Some of these computations occur in maps--arrays of neurons in which the tuning of neighboring neurons for a particular parameter value varies systematically. Computational maps transform the representation of information into a place-coded probability distribution that represents the computed values of parameters by sites of maximum relative activity. Numerous computational maps have been discovered, including visual maps of line orientation and direction of motion, auditory maps of amplitude spectrum and time interval, and motor maps of orienting movements. The construction of the auditory map of space is the most thoroughly understood: information about interaural delays and interaural intensity differences is processed in parallel by separate computational maps, and the outputs of these maps feed into a higher order processor that integrates sets of cues corresponding to sound source locations and creates a map of auditory space. Computational maps represent ranges of parameter values that are relevant to the animal, and may differentially magnify the representation of values that are of particular importance. The tuning of individual neurons for values of a mapped parameter is broad relative to the range of the map. Consequently, neurons throughout a large portion of a computational map are activated by any given stimulus, and precise information about the mapped parameter is coded by the locations of peak activity. There are a number of advantages of performing computations in maps. First, information is processed rapidly because the computations are preset and are executed in parallel. Second, maps simplify the schemes of connectivity required for processing and utilizing the information. Third, a common, mapped representation of the results of different kinds of computations allows the nervous system to employ a single strategy for reading the information. Finally, maps enable several classes of neuronal mechanisms to sharpen tuning in a manner not possible for information that is represented in a non-topographic code.

Keywords

ComputationComputer scienceComputational modelStimulus (psychology)Orientation (vector space)Representation (politics)Parameter spaceAuditory systemInterval (graph theory)AlgorithmArtificial intelligencePattern recognition (psychology)MathematicsNeuroscience

MeSH Terms

AnimalsAuditory CortexAuditory PathwaysBrain MappingInferior ColliculiOlivary NucleusPerceptionTime FactorsVisual CortexVisual Pathways

Affiliated Institutions

Related Publications

Publication Info

Year
1987
Type
review
Volume
10
Issue
1
Pages
41-65
Citations
385
Access
Closed

Social Impact

Altmetric

Social media, news, blog, policy document mentions

Citation Metrics

385
OpenAlex
7
Influential
316
CrossRef

Cite This

Eric I. Knudsen, Sophie Lac, SD Esterly (1987). Computational Maps in the Brain. Annual Review of Neuroscience , 10 (1) , 41-65. https://doi.org/10.1146/annurev.ne.10.030187.000353

Identifiers

DOI
10.1146/annurev.ne.10.030187.000353
PMID
3551761

Data Quality

Data completeness: 81%