The ventral midline thalamus coordinates prefrontal-hippocampal neural synchrony during vicarious trial and error

Abstract

When faced with difficult choices, the possible outcomes are considered through a process known as deliberation. In rats, deliberation is thought to be reflected by pause-and-reorienting behaviors, better known as vicarious trial and errors (VTEs). While VTEs are thought to require medial prefrontal cortex (mPFC) and dorsal hippocampal (dHPC) interactions, no empirical evidence has yet demonstrated such a dual requirement. The nucleus reuniens (Re) of the ventral midline thalamus is anatomically connected with both the mPFC and dHPC, is required for HPC-dependent spatial memory tasks, and is critical for mPFC-dHPC neural synchronization. Currently, it is unclear if, or how, the Re is involved in deliberation. Therefore, by examining the role of the Re on VTE behaviors, we can better understand the anatomical and physiological mechanisms supporting deliberation. Here, we examined the impact of Re suppression on VTE behaviors and mPFC-dHPC theta synchrony during asymptotic performance of a HPC-dependent delayed alternation (DA) task. Pharmacological suppression of the Re increased VTE behaviors that occurred with repetitive choice errors. These errors were best characterized as perseverative behaviors, in which some rats repeatedly selected a goal arm that previously yielded no reward. We then examined the impact of Re suppression on mPFC-dHPC theta synchrony during VTEs. We found that during VTEs, Re inactivation was associated with a reduction in mPFC-dHPC theta coherence and mPFC-to-dHPC theta directionality. Our findings suggest that the Re contributes to deliberation by coordinating mPFC-dHPC neural interactions.

Figure 1

Experimental design. (A) Delayed Alternation task schematic. Each trial requires the rat to select the goal arm opposite to the goal arm chosen on the previous trial to receive food reward. All trials, including the first trial of the session, are free-choice trials. A 30 s delay period separates trials, during which the rat is confined to the start-box. Green cup indicates reward, white cup indicates no reward. Blue arrow indicates correct trajectory, red arrow indicates incorrect trajectory. (B) During the baseline epoch, rats performed a set of 12–30 trials (DA Baseline epoch). They then received an infusion of saline, infusion of muscimol, or no infusion and returned to their home cage. Thirty minutes later, rats were tested on another set of 12–20 trials (DA Testing Epoch; see Hallock et al.). (C) Schematic of recording sites in the mPFC and dHPC and with a cannula site in the Re/Rh. (D) Histological confirmations of recording and cannulae placements. Colored dots in the mPFC (left) indicate different tetrode placements (see Hallock et al.).

Figure 2

Re inactivation disrupts choice-accuracy during both VTE and non-VTE. (A) Re inactivation did not change the percentage of trials with VTE. (B) Choice accuracy on trials with VTE, as measured by % correct, was significantly different between control and muscimol groups. Paired t-test with a significance level of 0.05. “Control” refers to collapsing across all sessions except muscimol testing (Supplementary Fig. S3). (C) Re inactivation disrupted choice accuracy during non-VTE trials as measured via repeated measures ANOVA, then with t-tests against a null of 0 (significance threshold is 0.0167). *p < 0.05. **p < 0.01. ***p < 0.001. Data are displayed as the mean ± s.e.m.

Figure 3

Re inactivation increased VTEs on choice error sequences. (A) An example rat that exhibited VTEs on every trial following Re inactivation, with > 50% of trials being perseverations. Left panels display position data from an example baseline (gray) and muscimol testing (red) trial. Overlaid are example trajectories where warm colors indicate greater normalized head velocity. Bar graphs (right) demonstrate that this example rat exhibited both perseverative and deliberative choice behaviors during muscimol testing (red) which were less prevalent across the various control conditions (gray) (see Rat 1 Fig. 2A). (B) Percentage of perseveration trials that were VTEs or non-VTEs were used to generate a normalized difference score between control and muscimol sessions. Notice that Re inactivation caused a significant increase to perseveration trials with VTEs but not to perseveration trials with non-VTEs. (C) In general, % perseveration was increased by Re inactivation. (D) Unlike perseveration, which makes no assumption of spatial bias, turn bias reflects a difference between the amount of left and right choices (e.g. |#left–#right|/#trials). Re inactivation did not lead to a reliable increase in turn-bias as measured via one-way repeated measures ANOVA. Notice that the numerical change in turn bias is best explained by the data in (C). Data are displayed as the mean ± s.e.m. **p < 0.01 ***p < 0.001.

Figure 4

Re inactivation disrupts mPFC-dHPC theta synchrony specifically during VTEs. (A) Maze schematic demonstrating that LFP data was extracted from the choice point (blue box surrounding T-junction). Middle panel demonstrates raw LFP. Right panel shows filtered LFP as a conceptual demonstration of theta phase coherence. (B) “Theta” was defined as a 5–10 Hz oscillation based on the power spectra from mPFC and HPC LFP. (C) Frequency × Coherence plots demonstrating a clear reduction in theta coherence on VTEs (left panel) but not non-VTEs (right panel) after Re inactivation. Red colors indicate data from the muscimol testing session, while gray colors indicate data obtained across control sessions. (D) Normalized theta coherence was averaged across the 5–10 Hz range, then statistically compared between control sessions and the muscimol testing session (N = rats). There was a significant reduction in mPFC-dHPC theta coherence during VTE (left panel) but not non-VTE (right panel) trials (paired t-tests, significance threshold of 0.025 for two tests). (E) There was no significant difference in time spent at the choice point during VTEs, although the p-value was “trending”. (F) Pearson’s correlation was performed between time spent at the choice point and mPFC-dHPC theta coherence. Three separate analyses were performed to isolate control (gray) and muscimol testing (red) datasets (N = 7 rats per group), and then to combine these data (black). Regression lines are color coordinated accordingly. If reductions in mPFC-dHPC theta coherence (D) were being driven by increased time spent at the choice point (E), then we should observe negative correlations between time spent and theta coherence. Bar graphs and shaded error bars are represented as the mean ± s.e.m. *p < 0.025.

Figure 5

Re inactivation reduces mPFC-to-dHPC theta directionality during VTEs. (A) Schematic representing LFP extracted from the choice point (left) and a conceptual representation of Granger prediction (right panels). Notice that Granger prediction provides estimates of predictive power in both directions (e.g. PFC-to-HPC and HPC-to-PFC). (B) Re suppression reduced mPFC-to-dHPC theta directionality during VTEs (N = 7) when compared to non-VTEs (N = 6). Note that 5 rats exhibited both VTEs and non-VTEs after reaching inclusion criteria for LFP analyses (see “Results” and “Methods”). (C) Re suppression did not disrupt dHPC theta leading mPFC theta. *p < 0.0167. Data are displayed as the mean ± the s.e.m.