Volume 96, Issue 5, Pages 1192-1203.e4 (December 2017) Lateral Orbitofrontal Inactivation Dissociates Devaluation-Sensitive Behavior and Economic Choice  Matthew P.H. Gardner, Jessica S. Conroy, Michael H. Shaham, Clay V. Styer, Geoffrey Schoenbaum  Neuron  Volume 96, Issue 5, Pages 1192-1203.e4 (December 2017) DOI: 10.1016/j.neuron.2017.10.026 Copyright © 2017 Terms and Conditions

Figure 1 Economic Choice Behavior in Rats (A) Illustration of a single trial from the choice task. A white noise cue indicates the availability to initiate a trial by nosepoking within a central port. Cues representing the current offer emerge on both screens after a 1 s hold which is followed by another 1 s hold before the white noise is turned off, indicating that the animal can touch one of the screens in order to make a choice. The general shape of the symbol indicates the good being offered (i.e., vertical bars and crescents), whereas the number of segmentations of the symbol indicates the number of pellets available for that good. (B) Representative examples of choice behavior. Each row of choice curves are taken from consecutive sessions of a single rat performing the three possible pairings of the goods listed. The goods are arranged in preference order (A > B > C) and are determined by the three sessions. x axis shows the offers given in log scale, and y axis shows the percent of the less preferred good chosen for each offer type. Indifference points (IP, shown as X = Y∗IP) are defined by a probit regression (solid black line) crossing the 50% choice value (horizontal dashed line). (C) All sets (n = 54) of three consecutive sessions comparing the indifference points of three goods pairs. Each point represents the indifference point of pair A:C (y axis) and pairs A:B∗B:C (x axis) with ideal transitive behavior falling on the line of equality (dotted line). Solid black line, OLS regression. Blue marker, example behavior from (B). (D) Reaction times are slower for offers close to the indifference point. Offers are binned relative to the indifference point of a session (x axis) and average reaction time is in seconds (y axis, average of all sessions shown in (C). Error bars indicate SEM. See Figure S1 for all visual cues used and Figure S2 for additional examples of choice behavior. Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions

Figure 2 Inactivation of OFC Using Halorhodopsin during the Cue Period of the Choice Task (A) Design for the inactivation experiment. (B) Histological verification of viral expression (top) and fiber placement (bottom) for halo and eYFP groups (left and middle) at ∼3.0 mm anterior of bregma. Example of eYFP expression (green) and DAPI (blue) (right, top) with an expansion (right, bottom) of the highlighted region (red). (C) Typical examples of the behavior observed for rats without expression of halorhodopsin (eYFP group, top) and rats with expression of halorhodopsin (Halo group, bottom) in the OFC. Behavior is broken down by fiber type (left, blocked; right, patent) and trials with which the laser was on or off (colored/gray curves). The indifference point (IP) is shown for each session. (D) Average behavior of all sessions on the choice task (eYFP group, n = 46 top; Halo group, n = 56 bottom) for blocked and patent fibers (left/right) and for laser-on (colored curves) and laser-off (gray curves) trials. Note that the behavior from each session was realigned to the IP of that session, and that the two curves, laser-on and laser-off, are overlapping one another in most panels. The gray dotted line shows the shift in behavior expected if all sessions shifted to an IP of 1:1. Error bars indicate SEM. Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions

Figure 3 No Effect of OFC Inactivation on the Indifference Point (A) Indifference points (IPs) from 46 sessions from the eYFP group (n = 7, top) and 56 sessions from the Halo group (n = 9; bottom) are shown for the blocked and patent patch cable fibers (left/right), split by laser-on trials (colored histograms) and laser-off trials (gray histograms). (B) Scatterplot of the paired within-session differences between laser-on and laser-off trials for each session-type (blocked and patent fibers, x and y axis, respectively) bounded by corresponding histograms on each axis. (C) Boxplots of IPs for each treatment corresponding to the histograms in (A). (D) The average difference in IP for laser-on and -off trials calculated only using sessions with IPs greater than the IP threshold (x axis) for the Halo group (left) and the eYFP group (right). For this analysis sessions close to 1:1 were progressively removed to up to 90% of the total sessions analyzed. Gray vertical lines in (D) show the IP threshold corresponding to a significant IP using a bootstrap method (see methods). All axes representing indifference points are plotted in log scale. Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions

Figure 4 No Effect of OFC Inactivation on the Inverse Slope (σ) for Individual Sessions (A) The inverse slope (s, estimated sigma from probit regression) from 46 sessions from the eYFP group (n = 7, top) and 56 sessions from the Halo group (n = 9, bottom) are shown for the blocked and patent patch cable fibers (left/right), split by laser-on trials (colored histograms) and laser-off trials (gray histograms). (B) Scatterplot of the paired within-session differences between laser-on and laser-off trials for each session-type (blocked and patent fibers, x and y axis, respectively) bounded by corresponding histograms on each axis. (C) Boxplots of the inverse slope for each treatment corresponding to the histograms in (A). (D) The average difference in inverse slope for laser-on and -off trials calculated only using sessions with IPs greater than the IP threshold (x axis) for the Halo group (left) and the eYFP group (right). For this analysis sessions close to 1:1 were progressively removed to up to 90% of the total sessions analyzed. Gray vertical lines in (D) show the IP threshold corresponding to a significant IP using a bootstrap method (see STAR Methods). Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions

Figure 5 Transitivity Is Unaffected by OFC Inactivation (A) Transitivity plots are shown for the eYFP group (top) and the Halo group (bottom), with patch cord fiber types split by blocked (left) and patent (right). The y and x axes show the IP, in log scale, for the session with the largest IP across pairs (A-C) and the product of the IPs from the other two sessions (A and B, B and C), respectively. The within-session pairs of laser-on trials (colored symbols) and laser-off trials (gray symbols) are indicated by a dark gray line connecting pairs of the symbols. Histograms comparing laser-on and laser-off trials are shown in the background for each of the plots. (B) The cumulative empirical distribution function of the distances (dark gray lines in A). Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions

Figure 6 Confirmation of the Efficacy of OFC Inactivation Using a Pavlovian Devaluation Task In order to determine whether OFC was indeed being inactivated by laser stimulation of halorhodopsin, a simple Pavlovian reinforcer devaluation procedure was performed which has been well established to be OFC dependent. (A) The same rats that were used in the choice task underwent 8 days of Pavlovian training using a 10 s auditory cue resulting in food delivery as well as a second auditory cue resulting in no food delivery. Food cup responding was measured over the final 5 s of the cue and reported as mean SEM. Initial training was followed by conditioned taste aversion (lithium chloride) to the food pellet used as the reinforcer (B). Preference tests for the devalued pellet as well as a control non-devalued pellet are shown before (B, top right) and after (B, bottom right) the devaluation procedure. The rats were then tested in an extinction probe test for a reduction in responding to the cue predicting the now devalued food pellet (C). The Halo group (n = 8, gold) failed to show a reduction in responding as compared to the eYFP group (n = 6, green) for both the cue period (left), during which the laser was on, and the post-cue period (right), during which the laser was off (10 s). All bars represent mean SEM. ∗p < 0.05 significance level corrected for multiple comparisons. Neuron 2017 96, 1192-1203.e4DOI: (10.1016/j.neuron.2017.10.026) Copyright © 2017 Terms and Conditions