Following d´ analyses, we next used MI (an information theoretic statistical approach; Cover and Thomas, 2005; Timme and Lapish, 2018) to precisely quantify the total amount of information encoded by each neuron. This approach is preferable to other parametric statistical analysis of firing rate, as firing rate distributions are highly non-normal (Roxin et al., 2011; Timme et al., 2016). We focused these analyses on two categorical domains, the amount of information encoding real trials versus null trials (collectively referred to as trial-encoding), and the amount of information encoding drinking trials versus non-drinking trials (collectively referred to as drink-encoding). Null trials were constructed from periods of the neural recording that were randomly selected from the ITI such that full null trials did not overlap real trials at any time.
