Striatal Microstimulation Induces Persistent and Repetitive Negative Decision-Making Predicted by Striatal Beta-Band Oscillation

Abstract

Persistent thoughts inducing irrationally pessimistic and repetitive decisions are often symptoms of mood and anxiety disorders. Regional neural hyperactivities have been associated with these disorders, but it remains unclear whether there is a specific brain region causally involved in these persistent valuations. Here, we identified potential sources of such persistent states by microstimulating the striatum of macaques performing a task by which we could quantitatively estimate their subjective pessimistic states using their choices to accept or reject conflicting offers. We found that this microstimulation induced irrationally repetitive choices with negative evaluations. Local field potentials recorded in the same microstimulation sessions exhibited modulations of beta-band oscillatory activity that paralleled the persistent negative states influencing repetitive decisions. These findings demonstrate that local striatal zones can causally affect subjective states influencing persistent negative valuation and that abnormal beta-band oscillations can be associated with persistency in valuation accompanied by an anxiety-like state.

Keywords: approach-avoidance conflict; basal ganglia; beta oscillation; caudate nucleus; cost-benefit decisions; emotion; macaque; motivation; primate; value judgment.

Figure 1.. The Ap-Av Task and the CN Microstimulation Effects

(A) Task procedure and experimental blocks. During the cue period, the red and yellow horizontal bars, respectively signaling the offered amounts of reward and punishment, appeared on the monitor. The monkeys made a decision between acceptance and rejection of the combined offer, and reported this by choosing either of two targets (cross for acceptance; square for rejection) that appeared during the response period. The locations of the targets were alternated randomly. (B) Mean Ap-Av choices. (C) Mean reaction times.(D and E) Examples of negative (D) and positive (E) stimulation effects. Ap (blue cross) and Av (red square) choices in the Stim-off (left panels) and Stim-on (middle panels) blocks in the same session and difference between these blocks (right panels). Decision matrices are plotted in left panels for the two blocks. Black contours in the right panels represent significant effect zones (p < 0.05, Fisher’s exact test). Dashed and solid lines in these panels respectively denote the decision boundaries in the Stim-off and Stim-on blocks. (F) Distribution of the effective sites. Stimulation results mapped onto Nissl-stained coronal sections. Blue and red circles indicate sites at which microstimulation respectively induced an increase in Ap and Av. Black crosses indicate non-effective sites. The size of the circles indicates the size of the effects. Data from monkey S were projected onto those from monkey P. (G) Proportions of N units (top) and P units (bottom) (number of N and P units / total number of recorded units), shown for 1-mm bins around the negative effective sites. Asterisk indicates the bin in which the proportion of N units or P units was significantly different from that of units aggregated over all tracks. *p < 0.05 (Fisher’s exact test). See also Figure S1.

Figure 2.. CN Microstimulation Induced Persistent Change in State Affecting Value Evaluation

(A) Changes in decision (%ΔAv - %ΔAp) across post-stimulation sessions. Stimulation-induced changes in decision were measured by comparing the decisions in the Stim-off and Stim-on blocks, and long-lasting effect was measured by comparing those in the Stim-off and Follow-up blocks (left). In the preceding sessions (middle), stimulation-induced changes in decision (white circles) were not significant. Purple region denotes the 95% confidence interval, and the upper limit was 3.3%. Gray shading extending from middle to right panels indicates the 5% significance threshold. After this control period, microstimulation induced increases in Av (red) or Ap (blue) decisions, relative to the Stim-off block (black), and some of these effects continued into Extended sessions (right). White circles: Recovery blocks. (B) Example of consecutive daily sessions. Stimulation induced increases in Av choices during the stimulation session (left panels) and the first Extended session (middle panels), but not during the second Extended session (right panels). Top panels show choices plotted in decision matrix; bottom panels show changes in decision between Stim-off and designated blocks. Dashed and solid lines denote the decision boundaries in, respectively, Stim-off and other blocks. Each panel is illustrated as in Figure 1D. (C) Econometric modeling. With the choice behavior in a single block (left), we performed logistic regression to construct a choice model (right). (D) Stimulation effects on decision boundaries. Distribution of points of intersections of the decision boundary in the Stim-on blocks (left), shown as percentage of sessions of positive (blue), negative (red) and non-effective (gray) sites. Stimulation did not change the distribution of x-intercepts (bottom left), but induced significant changes in intersection with y = 100 for effective sessions (top left). Change in slope of the decision boundary in the negative (Neg), positive (Pos) and non-effective (Non) sessions, derived for 4 trial periods in the Stim-on blocks indicated by color scale at right (right). Black dashed line indicates the decision boundary in the Stim-off block. ***p < 0.001, NS: p > 0.05 (t-test). (E) Summary of the negative stimulation effect. Mean changes in CBR (left) and decision (right) are plotted with changes in each block of consecutive sessions. Error bars indicate SEM. (F) Diazepam administration blocked stimulation-induced increase in Av choices. Left: Cartoon of protocol. Right: Diazepam significantly suppressed the negative effects (two-tailed t-test). Each data point (circle) was overlaid with 95% confidence interval (pink) and the mean (black). See also Figures S2 and S3.

Figure 3.. CN Microstimulation Increased Repetitive Av Choices

(A) Ap (blue crosses) and Av (red squares) decisions in the Stim-off (left) and Stim-on (middle) blocks in the same session, plotted in the decision matrix. The gradation of red squares corresponds to the number of consecutive Av choices, as indicated by scale at right. Dashed and solid lines respectively denote the decision boundaries in the Stim-off and Stim-on blocks. Right panel shows the difference between Stim-off and Stim-on blocks, illustrated as in Figure 1D. (B) Decision sequences in the Stim-off (top) and Stim-on (middle) blocks. The bottom panel shows the sequence of the offered reward (red circles) and airpuff (yellow circles) sizes. The same sequence was used in the Stim-off and Stim-on blocks. (C) Comparison between the observed number of repetition in the Stim-on block and the expected number of repetition derived from the RR procedure. Circles represent the counts of m repetitions of Av choice observed in the choice sequence. Black and gray lines show, respectively, the mean and the 95% confidence limit of the distribution produced by the RR procedure. Mean z-statistic representing deviation from the model is shown at right.

Figure 4.. CN Microstimulation Induced Unusually Repetitive Choices, whereas pACC Microstimulation Did Not

(A) Mean deviation in repetitive Av choices, measured in z-statistic, representing the difference between the observed behavior and the prediction of the RR procedure during sessions with CN (left) or pACC (right) stimulation. X-axis indicates the size of changes in decision, calculated by the difference in positive and negative effects (%ΔAv − %ΔAp). Circles indicate the sessions in which the stimulation induced significant deviation from the RR prediction (p < 0.05, z-test), and gray crosses are non-significant sessions. Stars indicate the sessions shown in Figures 3 and S3. Data points beyond the range of the y-axis are shown at the ceiling. Bars at right show the distribution of the mean deviations. (B) Proportion of sessions that had significant deviation (p < 0.05, z-test) from the prediction of the RR procedure. Proportion in the CN stimulation experiments was significantly larger than that in the pACC experiments (Fisher’s exact test). See also Figures S4–S6.

Figure 5.. Beta Oscillations Represent Decision-Related Variables

(A) Example of power spectrum (top) and spectrogram (bottom) of task-related LFP activity. Red and blue lines in the top panel indicate the mean spectra (± SEM) for, respectively, the precue and cue periods. Yellow shading indicates the beta range (13–28 Hz). The gray line (solid: precue, dashed: cue) indicates the fitted pink noise baselines. (B) Example of baseline-subtracted power spectra (± SEM) of the cue-period LFP activity that exhibited choice selectivity modulation during a negative effective session. Red and blue lines indicate the mean spectra (± SEM) for upcoming Av (red) and Ap (blue) choices, respectively. For each block, z-tests were performed between spectra for Ap and Av. Yellow shading indicates the frequency range with significant differences between Ap and Av in the power spectra. *p < 0.05 (z-test, Bonferroni corrected). (C) Beta response matrices projected onto the first two dimensions of the MDS (MDS map). Each cross indicates individual channel. Color indicates the group to which the channel belongs (N group: red; P group: blue; and other groups: green, cyan and magenta). To examine the similarity of each beta response to the behavioral choice pattern, MDS was performed with the Ap and Av behavioral responses shown on the right. The position of behavioral Ap choice (top right matrix) was at the leftmost position, and that of Av choice (bottom right matrix) at the rightmost position in the MDS map, indicating that the principal coordinate value (i.e., PCV) represents Av-Ap tuning of each beta response. (D) Mean beta responses of N (top) and P (bottom) groups plotted in decision matrices. Dashed lines: decision boundaries. See also Figure S7.

Figure 6.. Properties of the Beta Responses That Belong to N and P Groups

(A) Time-course of the z-values (top, Wilcoxon rank-sum test) and power (bottom, mean ± SEM normalized by precue activity) of beta responses that discriminated upcoming Av (red) and Ap (blue) choices and that were categorized as N (left) or P (right) groups. Colored highlights on x-axis show significantly larger magnitudes for Av (red) or Ap (blue) choice (p < 0.05, two-tailed t-test). (B) Cumulative onset times at which beta responses discriminated upcoming choices, showing significantly earlier onset in N group (red) than in the others. *p < 0.05 (Kolmogorov-Smirnov test). (C) Proportions of N-group (top) and P-group (bottom) beta responses (number of beta in each group / total number of recorded beta), shown for 1-mm bins around the negative effective sites. Asterisk indicates the bin in which the proportion of N-group or P-group beta response was significantly different from that of beta responses aggregated over all tracks. **p < 0.01 (Fisher’s exact test). See also Figure S7.

Figure 7.. Changes of Beta Representation Coincide with Behavioral Effects of CN Microstimulation

(A) Mean (± SEM) change in decision frequencies (top) and PCVs of cue-period beta responses (bottom) for effective (left) and non-effective (right) sessions. Mean PCV in the Follow-up block of effective sessions was significantly larger than that in recording-only sessions (gray line), but not in non-effective sessions. ***p < 0.001 (two-tailed t-test). (B) Classifications of the 84 beta responses in effective sessions (top) and those of 90 beta responses in non-effective sessions (bottom). Filled circles: beta responses of N-f channels in effective sessions. Crosses: the others in effective sessions. Open circles: beta responses in non-effective sessions. Colors show beta responses categorized as N (red), P (blue), and other (green, cyan and magenta) groups in the block indicated above each plot. (C) Proportion of N group for each block of effective (left) and non-effective (right) sessions, with significant increase in the Follow-up block relative to the proportion of N group in recording-only sessions (gray line). ***p < 0.001 (Fisher’s exact test). See also Figure S8.

Figure 8.. Modulation of Power Spectra in Abnormally Repetitive Sessions

(A) Precue power spectra of 16 channels that showed significant elevation for previous Av choice. Top: Mean (± SEM) power spectra for previous Ap (blue) and for previous Av (pink) choices. Bottom: Mean (± SEM) tuning indices derived by subtracting spectrum for previous Av from that for previous Ap spectra (black). Yellow shading indicates significant increase in the tuning index from that in the Stim-off block (gray). *p < 0.05 (t-test, Bonferroni corrected). (B) Precue power spectra of 7 channels that showed significant elevation for previous Ap choice.