- Poster presentation
- Open Access
Noise signature on interval timing
BMC Neuroscience volume 16, Article number: P180 (2015)
Among other essential adaptations is the capability of organisms to estimate durations in the seconds-to-hours range (interval timing). Such capabilities are critical for fundamental cognitive processes like decision making, rate calculation, and planning of action . In their seminal work on computational modeling of interval timing, Matell and Meck  revitalized the striatal beat frequency (SBF) model that utilizes the coincident activation of a series of oscillators to code for different durations. They showed through numerical simulations that the SBF model is capable of reproducing two of the interval timing signatures: (a) precise timing, i.e. the model output peaks at the training/criterion time and (b) scalar timing, i.e. the error in timing increases linearly with the criterion time. The SBF model was capable of producing scalar timing only when different types of biologically realistic variances (frequency, memory, etc.) were considered .
In this work, we investigated what effect each type of variance/noise has on the shape of the SBF model output. In particular, we noticed that the experimentally measured response rate is not quite Gaussian (see Figure 1A) and instead has a long tail. Mathematically, the output function of set of coincidental (sinusoidal) oscillators is given by :
where T is the criterion time, f i are the frequencies of neural oscillators. The criterion time T is learned during the training phase and stored/retrieved from the long-term memory with some errors. The firing frequencies f i of all neural oscillators also fluctuate. We found that memory variance (s T ) preserves the Gaussian shape of the output function, whereas the frequency variance (s f ) skews and has a long tail similar to experimental observations (see Figure 1B).
In addition to the significantly difference contributions to the shape of output functions, we also found that memory and frequency noises shift the peak of the Gaussian. Memory noise affects the storage/retrieval of criterion T and shifts the peak of the output to the right, i.e. t T = T(1+ T ), where T is a number that depends on the range of stored values of T and the probability distribution function (pdf T ) of the noise. Frequency variance shifts the peak of the output to the left, i.e. t f = T/(1+ f ), where f is a number that depended on the range of frequencies and the pdf f of the noise.
Gallistel CR: The organization of behavior. 1990, Cambridge, MA: MIT Press
Matell MS, Meck WH: Cortico-striatal circuits and interval timing: coincidence detection of oscillatory processes. Cognitive Brain Research. 2004, 21 (2): 139-170.
Oprisan SA, Buhusi CV: What is all the noise about in interval timing?. Philosophical Transactions of the Royal Society of London B: Biological Sciences. 2014, 369 (1637): 20120459-
Buhusi CV, Aziz D, Winslow D, Carter RE, Swearingen JE, Buhusi MC: Interval Timing Accuracy and Scalar Timing in C57BL/6 Mice. Behavioral Neuroscience. 2009, 123 (5): 1102-
This work was supported by the grant IOS CAREER 1054914 from the National Science Foundation.
About this article
Cite this article
Oprisan, S.A., Novo, D.N. Noise signature on interval timing. BMC Neurosci 16, P180 (2015). https://0-doi-org.brum.beds.ac.uk/10.1186/1471-2202-16-S1-P180
- Model Output
- Output Function
- Probability Distribution Function
- Frequency Variance
- Criterion Time