Probing the decision-making mechanisms underlying choice between drug and nondrug rewards in rats
- Lausanne University Hospital, Department of Psychiatry, Switzerland
- Université de Bordeaux, Institut des Maladies Neurodégénératives, France
- CNRS, Institut des Maladies Neurodégénératives, France
Figures
Figure 1 with 1 supplement
Experimental conditions in experiments included in the W/and W/O training sets.
(A) Diagram of the choice procedure comprising four sampling trials followed by eight choice trials, and separated by 10 min inter-trial intervals. (B) Experimental timeline in data sets without (W/O) and with (W/) prior operant training before choice testing. (C and D) Mean (± SEM) percentage of completed trials (C) and mean (± SEM) choice latency (D) across choice sessions in the W/O Training (white circles) and W/Training (black circles) sets. In the W/training set, only the last five choice sessions comprising data for all 150 rats are represented.
Figure 1—figure supplement 1
Illustration of assumptions and predictions of the sequential choice model (SCM).
(A) In a deliberative decision-making model, individuals assign values to the different options and compare these values at the time of choice to select to most profitable option (top panel). Alternatively, the SCM assumes that individuals make no comparison between options to make a choice (bottom panel). Instead, decision-making mechanisms are adapted to sequential encounters, in which individuals decide to accept or reject single opportunities. (B) The SCM proposes that individuals assign independently to each option a subjective value based on the option’s profitability relative to background opportunities. This value is expressed as a probability to accept each option instead of pursuing the search (i.e. the response latency). During a choice between two options A and B, the tendency to respond for both options is compared to a threshold. The response reaching the threshold first wins the race and is selected, while the alternative response is aborted (top panel). This results in a cross-censorship between distributions of response latencies (bottom panel); only the fastest response is selected and produces a latency observation. (C) When options are presented sequentially during sampling trials, there is no cross-censorship and the distributions of latencies are entirely expressed. (D) However, when options are presented simultaneously during choice trials, the cross-censorship described in (B) leads to a shortening of latencies during choice compared to sampling trials; the preferred option A is more often selected because response latencies for this option are overall shorter (top panel) but because of the overlap in distributions of latencies, the least preferred option B can occasionally be selected (bottom panel). Choices of option A are less censored by longer responses for option B (blue gradient, top panel); thus, the expected shift toward shorter latencies is weaker for the preferred option. In contrast, choices of the least preferred option B are largely censored by responses for option A (pink gradient, bottom panel). Thus, the expected shortening of choice latencies is stronger for the least preferred option.
Figure 2
Lengthening of saccharin choice latencies compared to saccharin sampling latencies in the W/O training set.
(A) Correlation between the latency ratio and the percentage of cocaine choice. (B) Correlation between the winning latency and the percentage of cocaine choice. (C) Distribution of preference scores. Only SP rats (purple circles; N = 29) were considered in the analysis. (D) Distributions of sampling and choice latencies during saccharin trials in SP rats. Transparent solid lines represent distribution of latencies for individual rats. (E) Box plot of saccharin sampling and choice latencies. The box extends from the lower to the upper quartile values with a horizontal line at the median. The whiskers extend from the box at 1.5 times the interquartile range. Gray lines represent mean sampling and choice latencies of individual rats. **p<0.0001.
Figure 3
Sampling latencies correlate with preference and predict preference profiles.
(A) Correlation between the latency ratio and the percentage of cocaine choice. (B) Correlation between the winning latency and the percentage of cocaine choice. (C) Distribution of preference scores and assignment of preference profiles in individual rats. SP: Saccharin-preferring rats, purple circles; IND: Indifferent rats, green circles; CP: Cocaine-preferring rats, blue circles. (D) The mean decoding accuracy (± standard deviation) of the preference profile, SP, IND, or CP, of individual rats based on their mean cocaine and saccharin sampling latencies (real – dark blue) is compared with the decoding accuracy expected from chance (shuffled – red; chance level 33.3% – horizontal dashed line). (E) Permutation test. The true difference between accuracy scores real-shuffled (vertical blue line) is compared to the distribution of differences in accuracy following permutations with 10,000 iterations. p<0.0001.
Figure 4 with 2 supplements
Shortening of cocaine choice latencies compared to cocaine sampling latencies.
(A–D) Distributions of sampling and choice latencies during saccharin trials in SP rats (A), cocaine trials in CP rats (B), and saccharin (C) or cocaine (D) trials in IND rats. Insets: Box plots of sampling and choice latencies. Boxes extend from the lower to the upper quartile values with a horizontal line at the median. The whiskers extend from the box at 1.5 times the interquartile range. Gray lines represent mean sampling and choice latencies of individual rats. *p<0.01, **p<0.0001.
Figure 4—figure supplement 1
Similar saccharin sampling and choice latencies when preference for saccharin is habitual.
(A and B) Box plots of sampling and choice latencies during the last three sessions preceding the first devaluation test using sensory-specific satiety (A) and the second devaluation test using conditioned taste aversion (B). Boxes extend from the lower to the upper quartile values with a horizontal line at the median. The whiskers extend from the box at 1.5 times the interquartile range. Gray lines represent mean sampling and choice latencies of individual rats. Red triangles represent the population mean.
Figure 4—figure supplement 2
Lengthening of latencies during choice compared to sampling when behavior becomes sensitive to devaluation.
(A and B) Box plots of sampling and choice latencies during privation sessions, before (A) and at the end (B) of devaluation training. Boxes extend from the lower to the upper quartile values with a horizontal line at the median. The whiskers extend from the box at 1.5 times the interquartile range. Gray lines represent mean sampling and choice latencies of individual rats. **p<0.01.
Tables
Table 1
Unique predictions of the different decision-making models.
| Models | Predictions |
|---|---|
| Deliberative choice model (DCM) | Sampling latencies < choice latencies |
| Sequential choice model (SCM) | Sampling latencies > choice latencies |
Table 2
Summary and conditions of experiments included in the analysis.
| N°Exp | Exp set | N | Prior training | Training session limit | Nb training sessions | Other conditions | Nb choice sessions (FR2) | Selected choice sessions |
|---|---|---|---|---|---|---|---|---|
| 1 | W/O training | 9 | N/A | N/A | 0 | *** | 5 | s13-s15 |
| 2 | W/O training | 11 | N/A | N/A | 0 | *** | 5 | s13-s15 |
| 3 | W/O training | 11 | N/A | N/A | 0 | *** | 5 | s13-s15 |
| Total | 31 | |||||||
| 4 | W/training | 18 | FR1 | 30 rewards/3 hr | 26 | *** | 8 | s32-s34 |
| 5 | W/training | 19 | FR1 | 30 rewards/3 hr | 26 | *** | 8 | s32-s34 |
| 6 | W/training | 11 | FR1 | 30 rewards/3 hr | 18 | 1 month home-cage saccharin access | 10 | s26-s28 |
| 7 | W/training | 22 | FR1/FR2 | 30 rewards/3 hr | 26 | cannula intra-OFC | 10 | s34-s36 |
| 8 | W/training | 21 | FR1 | 30 rewards/3 hr | 23 | cannula intra-OFC | 9 | s30-s32 |
| 9 | W/training | 12 | FR1/FR2 | 30 rewards/3 hr | 21 | *** | 9 | s28-s30 |
| 10 | W/training | 23 | FR1/FR2 | 30 rewards/3 hr | 32 | *** | 5 | s35-s37 |
| 11 | W/training | 9 | FR1 | 30 rewards/3 hr | 19 | *** | 5 | s22-s24 |
| 12 | W/training | 15 | FR1/FR2 | 20 rewards/2 hr | 21 | *** | 5 | s24-s26 |
| Total | 150 |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Probing the decision-making mechanisms underlying choice between drug and nondrug rewards in rats
eLife 10:e64993.
https://doi.org/10.7554/eLife.64993
