Task design and subject performance.

A) A trial begins when a rat pokes an illuminated port in the apparatus with its nose. Port engagement triggers initiation of a stream of Poisson-generated auditory pulses (“clicks”) at a baseline generative rate of 20 Hz. At a random point (marked by black arrow in this example) in 70% of trials, the generative rate increases by a variable magnitude of 10 Hz, 23 Hz, 36 Hz, 49 Hz. After the rate change, rats have 0.8 s to withdraw their nose from the initial port in order to receive a water reward from a second port. In the other 30% of trials, no change occurs, and rats must maintain port fixation for the full duration of the stimulus to receive a water reward. B) Combined hit rates as a function of change magnitude. Hit rate was calculated excluding false alarm trials (i.e. only including trials in which the rat was presented with a change). C) Reaction times for hits as the time of center port withdrawal following a change. D) Psychophysical reverse correlation (PRC). Calculated as the average local click rate preceding false alarms. Horizontal line shows the 20 Hz baseline rate. For all plots, error bars and error shading show SE. N = 4 rats.

Recording methods.

A) For each rat, a Neuropixels recording probe was implanted using a custom designed assembly that allowed use of a strain-relieving attachment for the tether. B) Probes were targeted stereotaxically to PPC (-3.8 to -3.95 mm AP, 2.2 mm ML).

Responses of PPC neurons positively modulated by evidence.

A) Population responses of neurons with significant positive modulation by evidence strength aligned to time of stimulus change (same rats as in Figure 1). Traces are normalized by the pre-change firing rates (from stimulus onset to change time) and sorted by change magnitude. Responses are plotted to the median reaction time for each change magnitude. Shading shows SE. N = 92 neurons. B) Population responses for hit trials of the same neurons aligned to the time of hit response. Traces are normalized by the pre-change firing rates and sorted by change magnitude. Shading shows SE. N = 92 neurons. C) Population click-triggered response of the same neurons. Residual z-scored responses aligned to each click are averaged across clicks for all neurons. Only clicks during the pre-change period are included, and neural responses are masked after the change (for hit and miss trials), movement response (for false alarm trials), or end of stimulus (for correct rejection trials). Shading shows SE. N = 92 neurons.

Behavioral and PPC neural responses for early and late stimulus periods.

A) Psychophysical reverse correlation (PRC) for the same rats as Figure 1, calculated as the average local click rate preceding false alarms and sorted by time of the false alarm. The blue trace corresponds to trials in which a false alarm occurred within 1.5 s of the stimulus start, and the red trace corresponds to trials in which a false alarm occurred later than 1.5 s after stimulus start. Horizontal line shows the 20 Hz baseline rate. Error shading shows SE. N = 3992 trials. B) Population click-triggered responses of the positively modulated neurons, sorted by time of the click. The blue trace corresponds to clicks presented within 1.0 s after stimulus start, and the red trace corresponds to clicks presented at least 1.0 s after stimulus start. For both, only clicks during the pre-change period are included, and all neural responses are masked after the change (for hit and miss trials), movement response (for false alarm trials), or end of stimulus (for correct rejection trials). Shading shows SE. N = 92 neurons. C) Click-triggered average (CTA) magnitude for each neuron comparing early clicks (y-axis) versus late clicks (x-axis), defined in the same manner as panel B. Each black point corresponds to one individual neuron. N = 92 neurons. The red point shows the average across neurons with error bars corresponding to the SE.

Responses of PPC neurons negatively modulated by evidence.

A) Population responses of neurons with significant negative modulation by evidence strength aligned to time of stimulus change (same rats as in Figure 1). Traces are normalized by the pre-change firing rates and sorted by change magnitude. Responses are plotted to the median reaction time for each change magnitude. Shading shows SE. N = 166 neurons. B) Population responses of the same neurons aligned to the time of hit response. Traces are normalized by the pre-change responses and sorted by change magnitude. Shading shows SE. N = 166 neurons. C) Population click-triggered response of the same neurons focusing on “late clicks” at least 1.0 second after stimulus start. Residual normalized responses aligned to each click are averaged across clicks for all neurons. Only clicks during the pre-change period are included, and neural responses are masked after the change (for hit and miss trials), movement response (for false alarm trials), or end of stimulus (for correct rejection trials). Shading shows SE. N = 166 neurons.

Rat behavioral performance following muscimol inactivation of PPC.

A) Hit rates as a function of change magnitudes comparing sessions with PPC infusion of vehicle (black) versus muscimol (magenta). Error bars show SE. B) False alarm rates during task performance following PPC infusion of vehicle (black) and muscimol (magenta). Error bars show SE. C) Correct trial reaction times for vehicle (black) and muscimol (magenta) sessions. D) Psychophysical reverse correlation (PRC) for vehicle (black) and muscimol (magenta) sessions. The horizontal line shows the 20 Hz baseline rate. Arrowheads show estimated start time of the increase in the PRC. Error shading shows SE. N = 3 rats for all panels.