Article Figures & Data
Figures
- Figure 1.
Sign tracker and goal tracker individuals differed mainly in their level of interaction with the lever cue. A, PCA index (see Materials and Methods) for all subjects measured during the last training session (day 7). Arrowhead, mean PCA index. Blue, subjects categorized as sign trackers; magenta, goal trackers. B–D, PCA index (B), total lever presses (C), and total magazine entries during the cue (D) over all 7 d of training for sign trackers (blue) and goal trackers (magenta). Error bars, SEM.
- Figure 2.
Histological reconstruction of recording locations in NAc core. Panels are coronal atlas sections (Paxinos and Watson, 2007) showing the location of electrode tips derived from electrolytic lesions and/or electrode tracks. Numbers are distance in millimeters from bregma.
- Figure 3.
The vigor of both ST and GT behavior may be represented in NAc firing. A, B, Example of a neuron with stronger cue-evoked excitation when the cue is followed by a lever press with shorter (A) rather than longer (B) latency. C, D, Example of a neuron with stronger cue-evoked excitation when the cue is followed by a magazine entry with shorter (C) rather than longer (D) latency. Left panels, action latency <50th percentile; right panels, action latency ≥50th percentile. Trials are shown in chronological order with earliest on top. Blue dots, cue onset; magenta triangles, magazine entries; cyan triangles, lever presses. E–G, On the population level, representation of latency to magazine entry (i.e., GT) predominates. E, Vigor index for latency to first action after cue onset. Median index is greater than 0.5 (p = 0.02, Wilcoxon). F, Vigor index for latency to magazine entry. Median index is greater than 0.5 (p < 0.001, Wilcoxon). G, Vigor index for latency to lever press. Distribution not different from 0.5 (p = 0.22, Wilcoxon signed-rank test). All panels, blue indicates significant vigor index (p < 0.05, permutation test); arrowhead indicates median. H, Average Spearman’s rank correlation coefficient (rho) between cue-evoked neural activity in the 500 ms following cue onset and the indicated behavioral measure for sign trackers (blue) and goal trackers (magenta). From left to right: number of lever presses, number of magazine entries, latency to first lever press, latency to first magazine entry. Single asterisk, p < 0.05, Wilcoxon rank-sum test; double asterisk, p < 0.001. Distribution of individual correlation coefficients may be found in Extended Data Fig. 3-1.
- Figure 4.
Sign trackers and goal trackers exhibit differences in NAc activity on the first day of training. A–D, Population average normalized activity aligned on cue onset (A, B) or reward delivery (C, D) for sign tracker (A, C) and goal tracker (B, D) subjects. Blue and magenta solid lines, first half of trials (trials 1–12); cyan and pink dashed lines, second half of trials (trials 13–25). Shading, SEM. E, F, Trial-by-trial normalized activity in response to reward delivery (1-s window) for sign trackers (E) and goal trackers (F). Error bars, SEM. G–J, Distribution of learning index for sign trackers (G, I) and goal trackers (H, J) derived from ROC analysis comparing the first half and second half of trials. Index >0.5 indicates higher cue-evoked (G, H) or reward-evoked (I, J) activity during early trials. Blue represents index significantly different from 0.5 (p < 0.05, permutation test). Arrowheads indicate median. The median is significantly greater than 0.5 for reward-evoked activity in sign trackers only (p < 0.001, Wilcoxon signed-rank test). K, PCA index for behavior from the final day of training plotted against the learning index for reward-related neural activity (1-s window). Regression line in red.
- Figure 5.
Sign trackers exhibit an attenuated reward response on the last day of training. A, B, Population average normalized activity aligned on cue onset (A) or reward delivery (B). Shading, SEM. Individual cell examples may be found in Extended Data Fig. 5-1. C, Population-wide average neural activity in the 1 s following cue onset (dashed lines) or reward delivery (solid lines) on the first day of training (left) and the last day of training (right). Error bars, SEM. All panels: sign trackers in blue, goal trackers in magenta.
- Figure 6.
In sign trackers, as compared to goal trackers, behavior and cue-evoked firing are resistant to extinction. A, B, Two example neurons recorded during the same extinction session. Within the same subject, some NAc neurons extinguish their cue-evoked firing (as in A), and some do not (as in B). Trials are shown chronologically with the earliest trial on top. Blue dots, cue onset. C, D, Population average activity during extinction sessions for extinguishing cells (C) and non-extinguishing cells (D). Shading, SEM. E, Cue-evoked neural responses in the 500 ms after cue onset for extinguishing cells (Ex.; solid lines) and non-extinguishing cells (N.Ex.; dashed lines). Activity is averaged in five-trial bins. F, G, Average behavior during extinction sessions for sign trackers (blue; lever presses only) and goal trackers (magenta; magazine entries only). The number (E) and latency (F) of actions are averaged in five-trial bins. Blue, sign trackers; magenta, goal trackers. All panels, dagger indicates p < 0.1, Wilcoxon rank-sum test. H, Average Spearman’s rank correlation coefficient (rho) between cue-evoked neural activity in the 500 ms following cue onset and the indicated behavioral measure for sign trackers (blue) and goal trackers (magenta). All comparisons between sign trackers and goal trackers are significant. From left to right: number of lever presses, number of magazine entries, latency to first lever press, latency to first magazine entry. Single asterisk, p < 0.05, Wilcoxon rank-sum test; double asterisk, p < 0.001. Distribution of individual correlation coefficients may be found in Extended Data Fig. 6-1.
Extended Data
Extended Data Figure 3-1
Correlation of the activity of individual neurons with trial-by-trial ST and GT behavior. Distribution of Spearman’s rank correlation coefficient (rho) relating cue-evoked neural activity (500-ms window) to number of lever presses (A, B), number of magazine entries (C, D), latency to first lever press (E, F), or latency to first magazine entry (G, H) for individual neurons recorded in sign trackers (A, C, E, G) or goal trackers (B, D, F, H) over the last 2 d of training. All panels, blue indicates significant correlation (α = 0.1), and p values indicate results of Wilcoxon signed-rank test for median different from zero. Download Figure 3-1, EPS file.
Extended Data Figure 5-1
Evolution of cue-related and reward-related activity over the course of training in two example neurons. Two representative neurons with highly stable waveforms over the course of training (seven sessions) were selected for analysis, one from a sign tracker (A, B) and one from a goal tracker (C, D). A, C, Heat plots show activity related to the cue (left) or reward (right) during each of the seven training sessions as average firing rates calculated in 10-ms bins with no smoothing. B, D, Average firing rate over the 500 ms following cue onset (dashed lines) or reward delivery (solid lines) for the cells shown in A, C, respectively. Error bars, SEM. Download Figure 5-1, EPS file.
Extended Data Figure 6-1
Correlation of the activity of individual neurons with behavioral extinction of ST and GT. Distribution of Spearman’s rank correlation coefficient (rho) relating cue-evoked neural activity (500-ms window) to number of lever presses (A, B), number of magazine entries (C, D), latency to first lever press (E, F), or latency to first magazine entry (G, H) for individual neurons recorded in sign trackers (A, C, E, G) or goal trackers (B, D, F, H) during an extinction session. All panels, blue indicates significant correlation (α = 0.1), and p values indicate results of Wilcoxon signed-rank test for median different from zero. Download Figure 6-1, EPS file.
In this issue
