Finally, because of the prominent role of place coding in views on hippocampal neural activity, we focused an additional analysis on a direct and quantitative comparison of the influence of time and location on neural

activity during the delay period; these analyses do not consider direction, speed, or their interactions. Using our GLM framework, we computed the likelihood of the data making use of only space selleck chemicals or time as the main variable and compared the outcome for each of those variables to that using a model that included the other variable as a covariate (see Experimental Procedures and Supplemental Experimental Procedures). These analyses indicated that time was informative in addition BI 2536 price to space for 131 (75%) of the delay neurons (χ25 > 11.1; p < 0.05). Similarly, for 138 (80%) of the delay neurons,

the addition of space augmented the amount of information already provided by time (χ25 > 11.1; p < 0.05). These proportions do not differ (χ21 = 0.6; p = 0.43). For 48 of the 175 delay neurons (27%), activity was best explained using only time or space because including both as covariates did not sufficiently improve the model. To evaluate whether space or time was more influential in these neurons, we compared the goodness of fit measure (i.e., the Akaike Information Criterion) obtained for a model that included only space or only time. For 20 out of these 48 neurons, time provided a better fit than a model that included only space, while space Cell press provided a better fit in the remaining 28 neurons. These proportions do not differ (χ21 = 2.04; p = 0.15). Nevertheless, for the majority of the delay neurons (127/175 or 73%), both space and time together provided significantly more information

than either variable by itself, suggesting that both influence their activity. For these neurons we also asked which dimension was more informative by defining a neuron’s spatiotemporal information content (STIC). For each neuron the STIC was computed by noting the increase in the likelihood of the model when one covariate—space or time—was added to a model that already included the other variable. The STIC was defined as positive when the addition of the time covariate to the space model was relatively more informative than the addition of the space covariate to the time model. Similarly, the STIC was negative when the opposite pattern was observed. The STIC of 67 neurons favored time, while that of 60 neurons favored space (Figure S3), and the mean of the distribution of STICs across the neuronal population did not differ from zero (0) (single-sample t test, H0: mean of the STIC = 0; t126 = 0.18; p = 0.86). Therefore, the population is equivalently influenced by both variables, and within the population the relative information provided by each dimension varies along a continuum.