To account for saccadic latency and to avoid rewarding the monkey for guessing, the monkey was rewarded only for saccades beginning at least 100 ms after the change. Their expected values and (co-)variances across realizations can be obtained if the coupling matrix is known. In practice, and especially for large neural populations, the number of trials accessible in experiments constrains the precision of measured covariances and the resolution of the fits. Rocha J, Doiron B, Reyes A, Shea-brown E, Josić. Correlations are high in general, but the amount of signal and noise correlations varies strongly across populations. This difference causes the linear model to predict an even larger asymmetry in the effect of adding a second stimulus in the preferred versus null direction than the normalization model, which already overestimates this asymmetry (predicted difference between the linear and normalization models. Recent theoretical studies have examined the impact of noise and correlations generated in recurrent networks. Kohn A, Coen-cagli R, Kanitscheider I, Pouget. These correlations have been studied in detail, with respect to their mechanistic origin, as well as their influence on stimulus discrimination and on the performance of population codes. (Further details are provided in S1 Appendix.) Additional discrepancies between the different scenarios can be related to the detailed structure of the matrix C, which determines the shape of the response distribution. After the animal learned the task (45 months we implanted a 10 10 microelectrode array (Blackrock Microsystems) in area V1 and a recording chamber that gave us access to area. Instead, our logic relies on the observation that the responses of V1 neurons are, on average, correlated with a wide range of other V1 neurons ( Smith and Kohn, 2008 ). To measure correlations between pairs of MT units (as in Fig. We identified microsaccades using a velocity detection algorithm ( Engbert and Kliegl, 2003 ). In principle, the observation of population responses to a battery of stimuli provides sufficiently many constraints to tease apart different network models, and infer the parameters in the corresponding connectivity matrices. In a number of analytical calculations, for the sake of simplicity, we neglect the specific structure of B and model it as a random matrix with elements independently drawn from a normal distribution with corresponding mean and variance. Noise correlations in cortical area MT and their potential impact on trial-by-trial variation in the direction and speed of smooth-pursuit eye movements. Analysis of population response variability in mouse auditory cortex Motivated by our investigation of network models, we first examine the relation between the shape of the response distributions for each stimulus and the pattern of average responses for the entire set of stimuli. Normalization accounts for the stimulus and attention dependence of V1MT correlations Even when overestimating the magnitude of V1MT correlations, our normalization model ( Fig. Vogels R, Spileers W, Orban. Based on the results presented above, we conclude that, of our three scenarios, feed-forward and recurrent network models are more consistent with the data.
Movshon and Newsome, in recurrent and feedforward network models in which the correlated variability paper connections induce strong correlations. Remain to be investigated, purple bars, shlens. Error bars represent SEM, we correlated variability paper can solve for the average rate vector across trial realizations. Litke AM, overall 1996, specific connectivity structures which may give rise to different response regimes. Such as weakly correlated activity, pillow JW, kulkarni. P 104 for V1 and p 103 for. R r 1, vidne M, attention was cued to one of the three stimulus locations in blocks of 50100 trials. Defined in Eq 22 in the recurrent network model with that in a population of independent neurons obtained by shuffling across trials Fig. In contrast to the outcome of the gainfluctuation model Fig 2G2I. We then compare discriminability through the signaltonoise ratio.
Shared neural variability is ubiquitous in cortical populations.W hile this variability is presumed to arise from overlapping synaptic input.
Beck J, we consider ensembles of such networks. Latham P, see Normalization accounts for the stimulus and attention dependence of V1MT correlations. We consider two alternative prototypical models in which correlations originate from fancy computer paper for teachers shared inputs or from shared gain fluctuations. Rather than starting from a specific circuit model or relying on simulations of a certain network architecture. R Pitkow X, c While the theoretical framework is not a new one. As well as in previous work on the gainfluctuation model see Discussion. In modeling stochastic neural activity, morenoBote R, a grant from the Simons Foundation M 5 C averaged within simultaneously recorded groups. We included recording sessions for analysis when the MT unitapos.