(A) Schematic peri-event time histograms (PETHs) illustrating putative responses to different cues under different hypotheses of how cue identity (light, sound; L, S) and outcome (reward-available, reward-unavailable; +, -) are coded. Left panel: Coding of identity is absent in the NAc. Top: Unit A encodes a motivationally relevant variable, such as expected outcome, similarly across other cue features, such as identity or physical location. Hypothetical plot is firing rate across time. L1+ (red) signifies a reward-available light cue, S1+ (navy blue) a reward-available sound cue, L2- (green) a reward-unavailable light cue, S2- (light blue) a reward-unavailable sound cue. Dashed line indicates onset of cue. Bottom: No units within the NAc discriminate their firing according to cue identity. Middle panel: Coding of identity occurs in a separate population of units from coding of other cue features such as expected outcome or physical location. Top: Same as left panel, with unit A discriminating between reward-available and reward-unavailable cues. Bottom: Unit B discriminates firing across stimulus modalities, depicted here as firing to light cues but not sound cues. Note lack of coding overlap in both units. Right panel: Coding of identity occurs in an overlapping population of cells with coding of other motivationally relevant variables. Hypothetical example demonstrating a unit that responds to reward-available cues, but firing rate is also modulated by the stimulus modality of the cue, firing most for the reward-available light cue. (B) Schematic PETHs illustrating potential ways in which identity coding may persist over time. Left panel: Cue-onset triggers a transient response to a unit that codes for cue identity. Dashed lines indicate time of a behavioural or environmental event. 'Cue-ON’ signifies cue-onset, 'NP’ signifies nosepoke at a reward receptacle, 'Out’ signifies when the outcome is revealed, 'OFF’ signifies cue-offset. Middle and right panel: Identity coding persists at other time points, shown here during a nosepoke hold period until outcome is revealed. Coding can either be maintained by a sequence of units (middle panel) or by the same unit as during cue-onset (right panel). (C) Schematic pool of NAc units, illustrating different analysis outcomes that discriminate between hypotheses. R values represent the correlation between sets of recoded regression coefficients (see text for analysis details). Left panel: Cue identity is not coded (A: left panel), or is only transiently represented in response to the cue (B: left panel). Middle panel: Negative correlation (r < 0) suggests that identity and outcome coding are represented by separate populations of units (A: middle panel), or identity coding is represented by distinct units across different points in a trial (B: middle panel). Red circles represents coding for one cue feature or point in time, blue circles for the other cue feature or point in time. Right panel: Identity and outcome coding (A: right panel), or identity coding at cue-onset and nosepoke (B: right panel) are represented by overlapping populations of units, shown here by the purple circles. The absence of a correlation (r = 0) suggests that the overlap of identity and outcome coding, or identity coding at cue-onset and nosepoke, is expected by chance and that the two cue features, or points in time, are coded by overlapping but independent populations from one another. A positive correlation (r > 0) implies a higher overlap than expected by chance, suggesting coding by a joint population. Note: The same logic applies to other aspects of the environment when the cue is presented, such as the physical location of the cue, as well as other time epochs within the task, such as when the animal receives feedback about an approach.