2018/02/06 by Agnieszka Pręgowska, Pregowska, Agnieszka, Ehud Kaplan +3 · 1 citation
Computer Science · Neuroscience · #60J20 #94A17 #94A24 #FOS: Biological sciences #Neural Networks and Applications #Neural dynamics and brain function #Quantitative Methods (q-bio.QM)
paper · pdf · doi:10.48550/arxiv.1802.01833
openalex publication_date 2018/02/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The nature of neural codes is central to neuroscience. Do neurons encode\ninformation through relatively slow changes in the emission rates of individual\nspikes (rate code), or by the precise timing of every spike (temporal codes)?\nHere we compare the loss of information due to correlations for these two\npossible neural codes. The essence of Shannon's definition of information is to\ncombine information with uncertainty: the higher the uncertainty of a given\nevent, the more information is conveyed by that event. Correlations can reduce\nuncertainty or the amount of information, but by how much? In this paper we\naddress this question by a direct comparison of the information per symbol\nconveyed by the words coming from a binary Markov source (temporal codes) with\nthe information per symbol coming from the corresponding Bernoulli source\n(uncorrelated, rate code source). In a previous paper we found that a crucial\nrole in the relation between Information Transmission Rates (ITR) and Firing\nRates is played by a parameter s, which is the sum of transitions probabilities\nfrom the no-spike-state to the spike-state and vice versa. It turned out that\nalso in this case a crucial role is played by the same parameter s. We found\nbounds of the quotient of ITRs for these sources, i.e. this quotient's minimal\nand maximal values. Next, making use of the entropy grouping axiom, we\ndetermined the loss of information in a Markov source in relation to its\ncorresponding Bernoulli source for a given length of word. Our results show\nthat in practical situations in the case of correlated signals the loss of\ninformation is relatively small, thus temporal codes, which are more\nenergetically efficient, can replace the rate code effectively. These phenomena\nwere confirmed by experiments.\n