Change-of-mind neuroeconomic decision-making is modulated by LINC00473 in medial prefrontal cortex in a sex-dependent manner

Abstract

Changing one's mind involves reappraisals between past costs versus future value and may be altered in psychopathology. Long intergenic noncoding RNA LINC00473 in medial prefrontal cortex (mPFC) can induce stress resilience in a sex-dependent manner, but its role in cognition is unknown. We characterized decision-making behavior in male and female mice in the neuroeconomic paradigm Restaurant Row following virus-mediated expression of LINC00473 in mPFC. Mice foraged for food among varying temporal costs and subjective value while on a limited time budget. Without affecting primary deliberative decisions, LINC00473 selectively influenced re-evaluative choices in a sex-dependent manner. This included changing how mice (i) cached value with the passage of time and (ii) weighed prior mistakes, which underlie the computational bases of sensitivity to sunk costs and regret. These findings suggest that a common value function is shared between these neuroeconomic processes and reveal a bridge between molecular drivers of stress resilience and psychological mechanisms underlying sex-specific proclivities in negative rumination.

Fig. 1.. Male and female mice similarly acquire the Restaurant Row task.

(A) Timeline. Representative images of surgical targeting and virus transfection of GFP at 40× and 10× magnifications with 4′,6-diamidino-2-phenylindole (DAPI) and NeuN staining. The mice were allowed to recover before beginning longitudinal testing on the Restaurant Row task for 45 consecutive days. Dashed box indicates time period relevant for this figure: week 1 of testing. Mouse artwork credit:https://scidraw.io/, L. Petrucco; https://doi.org/10.5281/zenodo.3925903 (https://creativecommons.org/licenses/by/4.0/). (B) Task schematic. The mice were allotted 30 min daily to invest time foraging for their primary source of food in a self-paced manner. Costs to obtain rewards were in the form of delays mice would have to wait near feeder sites. The mice were required to run in a counterclockwise direction encountering offers for different flavors at each “restaurant” in serial order. Each restaurant, separated by hallways, was divided into a T-shaped “OZ” choice point and a separate “WZ” that housed the pellet dispenser. Upon OZ entry from the correct heading direction, a tone sounded whose pitch indicated the delay the mice would have to wait if accepting the offer by entering the WZ. If entered, the tone pitch descended in the WZ, cuing the indicated delay. Each trial terminated if the mice skipped in the OZ, quit during the countdown in the WZ, or earned a reward, after which the animals were required to proceed to the next restaurant. Mouse artwork credit: https://scidraw.io/, F. Claudi; https://doi.org/10.5281/zenodo.3925997 (https://creativecommons.org/licenses/by/4.0/). (C to F) Behavioral metrics across the first week of testing during which all offers were 1 s only (lowest pitch, 4 kHz): (C) distance traveled, (D) laps run in the correct direction, (E) rewards earned, (F) earnings split by flavors ranked from least to most preferred by summing each day’s end-of-session totals in each restaurant. No effect of sex or LINC00473 expression was grossly apparent in these initial metrics. Error bars represent ±1 SEM.

Fig. 2.. LINC00473 expression in mPFC alters sex- and value-dependent, task-driven CPP behavior.

(A) Timeline. Dashed box indicates time period relevant for this figure: two special testing days marked with cyan arrows. (B) Experimental design. Animals were placed in the arena free to roam for 20 min on day 0 and day 8 of the experimental timeline but with no active task to obtain pre-task baseline and post-task experience–related exploratory behavior. (C) Total number of entries and total time spent in the WZ, aggregated in all restaurants. Mouse artwork credit: https://scidraw.io/, F. Claudi; https://doi.org/10.5281/zenodo.3925997 (https://creativecommons.org/licenses/by/4.0/). (D and E) Representative video tracking data from an example mouse (GFP-treated female) recorded during the pre (day 0) and post (day 8) tests. (D) Track plots displaying body position when traveling between restaurants (gray) and when in the WZ of each restaurant (colored). (E) Occupancy heatmaps of the same example mouse in (D). The change in entries into and time spent in each restaurant’s WZ can be calculated as a delta score (post minus pre); color of these example heatmaps is scaled across sessions to the maximum occupancy (130 s). (F and G) Delta scores in change of (F) total entries and (G) total time comparing post (day 8) minus pre (day 0) time points split by restaurants ranked according to rewards earned on the active task on day 7. Only the most preferred (1st) and least preferred (4th, faint) restaurants are depicted here (see fig. S4 for full dataset and replication cohort). Horizontal dashed gray line indicates delta score of 0. LINC00473 expression in mPFC abolished CPP behavior in most preferred restaurants only in females and only in the time but not entry domain. *Represent significant sex differences. Error bars represent ±1 SEM.

Fig. 3.. Escalating reward scarcity unmasks sex-dependent effects of LINC00473 expression in mPFC on complex foraging profiles.

(A) Timeline. Dashed box indicates time period relevant for this figure: entire 45-day Restaurant Row paradigm. (B) Rewards earned (black) and laps run in the correct direction (gray). All 40 mice. Color bars along the x axis reflect the distribution of offers illustrated in (A). Vertical dashed green-yellow-red lines indicate transition points of the stepwise schedule and are reused throughout all other figures as a visual aid. Inset depicts “hysteresis” plot of laps against earns (epochs of days are color-coded). (C) Rewards earned split by flavors ranked from most preferred (1st) to least preferred (4th, faint). All 40 mice. (D) Change in rewards earned in each restaurant comparing 2 days at various color-coded transition points (delta scores calculated by subtracting the following: i, days 8 and 7 (yellow); ii, days 13 and 12 (orange); iii, days 18 and 17 (red); iv, days 45 and 18 (purple). All 40 mice (see fig. S5 for group comparisons). (E) Portion of total session time engaged in four major choice behaviors: OZ—enter (green) or skip (red); WZ—quit (magenta) or earn (blue). Rings split restaurants ranked from most (1st, inner ring) to least preferred (4th, outer ring). Each of the five pie charts represent data from each testing stage, with the early and late epochs during the 1- to 30-s offer range from the first 5 and last 5 days of that period. (F) Laps run in the correct direction split by sex and treatment. (G) Proportion of enter decisions terminating in a quit outcome once in the WZ. LINC00473 expression in mPFC drove females to run more laps in response to a worsening economic landscape and quit more frequently. *Represent significant sex differences. Shading/error bars represent ±1 SEM.

Fig. 4.. Change-of-mind decision-making patterns reveal unique sex-dependent neuroeconomic policies.

(A) Timeline. (B) Example behavior from a single mouse during the 1- to 30-s epoch (early, days 18 and 22; late, days 41 to 45). Example OZ and WZ thresholds indicated by the vertical dashed green (oz th) and blue (wz th) lines respectively in each restaurant. (C) Summary data from all 40 mice early or late during the 1- to 30-s epoch. (D) Thresholds of willingness to accept offers in the OZ or earn offers in the WZ. Inset depicts an example session from a single mouse in one restaurant. All 40 mice. Top inset depicts hysteresis plot of OZ against WZ thresholds (epochs color-coded). (E) OZ thresholds (top) and WZ thresholds (bottom) split by sex and treatment. (F) Thresholds in (E) from early and late 1- to 30-s epoch. (G to I) Economic analysis of how decision policies during the 1- to 30-s epoch approximate maximal food versus subjective value (see Methods and fig. S7 for full explanation). (G) Individual differences in (top) SD of rewards earned across flavors on day 7, (middle) total rewards earned on day 7, and (bottom) earns split by restaurant (flavor: chocolate, brown; banana, yellow; grape, purple; and plain, cyan) and size (reward count) on day 7. All data in (G) are sorted by SD data in (top). Small colored boxes along the x axis indicate sex (male, blue; female, red) and treatment (LINC, dark; GFP, light). (H) Example data from a single mouse on day 7 showing earn ratios between flavors. (I) Data from (G) and (H) used in a computational modeling analysis to calculate the Euclidean distance between observed and theoretical decision policies for maximal food (brown) or subjective value (magenta) separately for OZ and WZ thresholds. *Represent significant sex differences. Shading / error bars represent ±1 SEM.

Fig. 5.. LINC00473 expression in mPFC alters the temporal dynamics of change-of-mind decisions.

(A) Distribution of quit trials by offer value or value left from the 1- to 30-s epoch. (B) Average value left at quit. (C) Three categories of quit events based on offer value and value left. (D) Number of quit events from each quit category in (C). (E) Split by restaurants ranked from least (4th) to most preferred (1st) from the 1- to 30-s epoch. (F to H) Sunk cost analysis of staying behavior in the WZ. (F) Likelihood of staying in the WZ as a function of time left in the countdown and time already spent waiting. Black 0 s time spent represents just entering the WZ. Inset illustrates an example analysis point matched at 10 s left. Data from (F) dimensioned reduced in (G) collapsing across time left. Insets in (G) depict data from curves in (F) collapsed into the observed (sunk condition) and control (0 s condition) lines. Difference between curves in (G) is plotted in (H) to summarize the envelope of the overall effect of time already spent on staying in the WZ. (I) Sensitivity to sunk costs across the entire paradigm. (J and K) Regret-related sequence analyses comparing trial pairs between (J) type 1 scenarios and (K) type 2 scenarios. Read-out trials captures a change in behavior depending on choice history of distinct economic violations (black) compared to nonviolations (gray) on trial t-1. See Methods for a full explanation of these analyses. Prominent sex-dependent effect revealed by LINC00473 expression in mPFC on change-of-mind–related sunk costs (enhanced in females) and regret sensitivity (type 2: enhanced in females). * in (B) to (I) represent significant sex differences. * in (J) and (K) represent delta score significantly different from 0. Shading/error bars represent ±1 SEM.

Fig. 6.. Summary of key findings.

Distinction between decision-making processes measured within trial in the OZ (deliberative choices between skipping versus entering) and WZ (re-evaluative choices between quitting versus continuing to wait before earning a reward, star symbol) as a function of cued delay costs (tone pitch, colored notes) and subjective value (flavor). Analysis of foraging behaviors reveal signatures of dissociable neuroeconomic processes in the OZ (willingness to enter, vicarious trial and error choice behavior, behavioral sensitivity to type 1 regret scenarios: the impact of skipping high-value offers on subsequent choices) and WZ (willingness to wait, behavioral sensitivity to sunk costs, and behavioral sensitivity to type 2 regret scenarios: the impact of entering and then quitting low-value offers on subsequent choices). Interlocking male and female symbols (⚤) represent no sex differences, while separated symbols (♀ ♂) indicate sex differences, either at baseline (GFP group) or altered by LINC00473 expression in mPFC. See fig. S14 for an expanded summary table. Mouse artwork credit: https://scidraw.io/, F. Claudi; https://doi.org/10.5281/zenodo.3925997 (https://creativecommons.org/licenses/by/4.0/).

Fig. 7.. Conceptual neuroeconomic framework.

Working model of the link between key WZ phenotypes enhanced in LINC00473-treated females during change-of-mind decision-making. This schematic depicts the choice conflict after mice accept a high-cost offer—an economic violation in the OZ—once they arrive in the WZ and are faced with competing actions to stay versus quit. Clock emphasizes that this decision process evolves with the passage of time. Quitting rapidly is most economically advantageous. Yet, accruing sunk costs promotes staying as added value is cached over time. If these trials are quit—a defining feature of type 2 regret scenarios—this added value is carried into subsequent trials, augmenting future choices. Red arrows highlight enhanced sensitivity to sunk costs, and type 2 regret apparent in LINC00473-treated females compared to baseline (gray). Sunk costs: p(stay) increases as a function of time spent waiting on trial T; type 2 regret: p(compensate) (increased likelihood of accepting offers typically rejected on trial T + 1) is higher following violation sequences (enter-then-quit) relative to nonviolation control sequences (skip) for high-cost offers on trial T. Proposed unifying principle linking sensitivity to sunk costs and type 2 regret: Both phenomena stem from the same trial pool on trial T: inappropriately accepting a high-cost offer. Alterations in reward value, driving either sunk cost–induced staying or post-quit regret–related compensatory behavior, may stem from a shared value function. There may be a common underlying counterfactual process where the unselected alternative (i.e., skipping trial T instead) carries added value and influences behavior either during staying or after quitting. This value function in either case minimizes future loss. This work makes explicit predictions about hypotheses for future research investigating the neural correlates of shared counterfactual representations and, clinically, if the content of negative rumination in mood disorders maps onto these representations. Mouse artwork credit: https://scidraw.io/, A. Park; https://zenodo.org/records/10940481 (https://creativecommons.org/licenses/by/4.0/).