Rough-and-tumble play or play fighting is an important experience in the juvenile period of many species of mammals, as it facilitates the development of social skills, and for some species, play fighting is retained into adulthood as a tool for assessing and managing social relationships. Laboratory rats have been a model species for studying the neurobiology of play fighting and its key developmental and social functions. However, play fighting interactions are complex, involving competition and cooperation; therefore, no single measure to quantify this behavior is able to capture all its facets. Therefore, in this paper, we present a multilayered framework for scoring all the relevant facets of play that can be affected by experimental manipulations and the logic of how to match what is measured with the question being asked. © 2022 Wiley Periodicals LLC.

Many problem gamblers have comorbid depression and anxiety. We have previously shown that chronic exposure to gambling-like (random-ratio) schedules of reinforcement (compared to fixed-ratio schedules) was not enough to induce compulsive reward-seekingbehaviour in rats. Would an animal model of depression be more sensitive to random rewards? In our first experiment, we trained 12 Wistar-Kyoto (WKY) rats, an animal model of depression, and 12 Wistar (WIS) controls to press a lever multiple times to receive food rewards on a random-ratio schedule. Despite showing lower overall rates of lever pressing, WKY rats showed evidence of increased persistence. Namely, WKY rats were less sensitive to cues predicting reward unavailability and they persevered for much longer when reward was paired with increasing foot shock intensity. To assess if these effects were specific to random-ratio schedules, and hence could serve as a model of gambling addiction, we next tested 18 WKY and 18 WIS rats on random-ratio or fixed-ratio schedules. As in our first study, WKY rats showed perseverative tendencies, but did so equally under both random- and fixed-ratio schedules. This suggests that with WKY rats, and perhaps also with depressed gamblers, it is the propensity to get lost in the repetitive experience, rather than the unpredictability of reward, that makes these activities so alluring.

Rats emit a variety of calls in the 40–80 kHz range (50 kHz calls). While these calls are generally associated with positive affect, it is unclear whether certain calls might be used selectively in certain contexts. To examine this, we looked at ultrasonic calls in 30–40 day old male rats during the expectation of either play or food, both of which are reinforcing. Behavior and vocalizations were recorded while rats were in a test chamber awaiting the arrival of a play partner or food over seven days of testing. Control groups were included for the non-specific effects of food deprivation and social isolation. Play reward led to an increase in 50 kHz vocalizations, generally, with specific increases in trill and “trill with jump” calls not seen in other groups. Expectation of food reward did not lead to a significant increase in vocalizations of any type, perhaps due to the young age of our study group. Further, rats that were food deprived for the food expectation study showed markedly lower calls overall and had a different profile of call types compared to rats that were socially isolated. Taken together, the results suggest that trill-associated calls may be used selectively when rats are socially isolated and/or expecting a social encounter.

Conference presentations and research posters presented at the event made available with the permission of the authors.

The Wistar-Kyoto (WKY) rat was developed as a control for the spontaneous hypertensive rat but has subsequently also been used as a genetic animal model of depression due to its hyper-responsiveness to stress. We used anticipation of social reward (i.e., a play partner) to assess behavioural and vocal differences between the WKY and normal Wistar (WI) rats in the juvenile period. We found marked differences between groups; the WKY rats, were less active, vocalized less, and used significantly fewer types of 50-kHz calls in comparison to their WI counterparts. The animals were re-tested in adulthood and the same differences existed in overall activity, types of vocalizations and the behavioural vocal profiles used by the two groups of animals. These findings provide a robust baseline for an animal model of depression using a social paradigm. This paradigm may be useful to evaluate the efficacy of pharmaceutical interventions as potential treatments of depression in WKY rats.

Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Psychostimulants such as d-amphetamine (AMPH) often have behavioral effects that appear paradoxical within the framework of optimal choice theory. AMPH typically increases task engagement and the effort animals exert for reward, despite decreasing reward valuation. We investigated neural correlates of this phenomenon in the anterior cingulate cortex (ACC), a brain structure implicated in signaling cost-benefit utility. AMPH decreased signaling of reward, but not effort, in the ACC of freely-moving rats. Ensembles of simultaneously recorded neurons generated task-specific trajectories of neural activity encoding past, present, and future events. Low-dose AMPH contracted these trajectories and reduced their variance, whereas high-dose AMPH expanded both. We propose that under low-dose AMPH, increased network stability balances moderately increased excitability, which promotes accelerated unfolding of a neural ‘script’ for task execution, despite reduced reward valuation. Noise from excessive excitability at high doses overcomes stability enhancement to drive frequent deviation from the script, impairing task execution. Introduction Animals can draw from a large repertoire of innate and learned actions to generate behavioral output (Whishaw and Kolb, 2005). Even well-trained rodents sometimes engage in sleep, grooming, exploration, or other behaviors during laboratory tasks. When hungry, however, food rewards usually have sufficient value to motivate task engagement in lieu of these other options. Effort, reward, and other factors pertinent for decisions about resource collection are often formalized within the concept of utility (Phillips et al., 2007; Glimcher et al., 2009). Options requiring low effort and yielding a large food reward have high utility to hungry animals, whereas options requiring high effort or yielding little/unwanted food have low utility. Animals often seek to maximize utility, even if it requires exerting additional effort. For instance, rodents and primates typically choose to exert increased effort if the associated reward is of considerably higher value than that of lower-effort options (Salamone et al., 1994; Hosokawa et al., 2013). This preference is plastic. Moderate doses of d-amphetamine (AMPH) increases the amount of work rodents will exert for reward (Floresco et al., 2008b; Bardgett et al., 2009). It also increases engagement in learned food-seeking behaviors (Foltin, 2001; Odum and Shahan, 2004). The most obvious explanation for these effects according to utility theory is that AMPH increases the perceived value of reward. Behavioral data, however, suggest that reward valuation is decreased. AMPH-treated rats are less motivated to eat; their latency to first consumption is longer, they consume less than usual, and they spend less time eating (Blundell et al., 1976; Blundell et al., 1979; Leibowitz et al., 1986). Suppression of food-intake by AMPH is also evident in primates (Foltin, 2001). Why do animals work harder for food they appear less motivated to consume? It is possible that AMPH somehow affects the perceived cost of effort, or that increased task engagement is a by-product of motoric hyperactivity. But neither of these provide a robust explanation for the full suite of effects, including the drastic shift of responding to grooming and stereotyped outputs at high doses (Randrup et al., 1963; Randrup and Munkvad, 1967). The behavioral data, therefore, do not support a clear prediction about how AMPH may influence the neural encoding of utility. Cost-benefit decisions engage a network of brain structures, and the anterior cingulate cortex (ACC) appears particularly important for those involving physical effort (Rudebeck et al., 2006; Floresco et al., 2008a). The preference of rats for high-effort, high-reward options is reduced or eliminated by ACC lesions (Walton et al., 2002; Walton et al., 2003; Schweimer and Hauber, 2005; Holec et al., 2014). Electrophysiological recordings indicate that rat ACC and nearby regions in the medial prefrontal cortex (mPFC) encode a variety of signals related to choice and task execution. These include the position of the animal (Euston and McNaughton, 2006; Fujisawa et al., 2008; Mashhoori et al., 2018), task phase (Lapish et al., 2008; Balaguer-Ballester et al., 2011), reward (Gruber et al., 2010; Cowen et al., 2012), choice (Cowen et al., 2012), effort (Cowen et al., 2012; Hashemniayetorshizi et al., 2015), and other features (Cowen and McNaughton, 2007; Gruber et al., 2009; Durstewitz et al., 2010; Sul et al., 2010). Some ACC neurons also jointly encode costs-benefit information, which is well suited to utility signaling (Hillman and Bilkey, 2010; Cowen et al., 2012). These data are consistent with findings in monkeys and humans (Isomura et al., 2003; Kennerley et al., 2006; Croxson et al., 2009; Skvortsova et al., 2014; Blanchard et al., 2015; Klein-Flügge et al., 2016). Although AMPH clearly modulates ACC activity (Lapish et al., 2015), its effect on the encoding of effort-reward utility and other task variables in ACC has not been explicitly shown. Here, we attempt to link these independent observations to better understand how AMPH affects ACC encoding and dynamics pertinent to task engagement and outcome valuation. Results We used high-density electrophysiology to record ensembles of single neuron activity from the ACC of well-trained rats performing a continuous version of the classic T-maze (Figure 1A). Rats ran from a starting feeder to one of two target feeders accessible after turning to the right or left at the choice point. Rats were forced to alternate to the left and right sides on subsequent trials, and trials were organized into blocks in which either the reward volume delivered at the target feeders, or the effort (barrier climb) required to reach each target feeder, was different (Figure 1B). Rats performed five blocks of trials, then received an injection of either saline or AMPH, and performed the five blocks again. We quantified changes in behavior and neural signaling after the injection, with respect to the pre-injection phase. We recorded a total of 1209 putative pyramidal neurons from 22 session (55 ± 6.7 simultaneous cells per session) in four rats. Figure 1 Download asset Open asset Experimental apparatus and task. (A) Illustration of the figure 8 maze. The target feeders are elevated on some trials to impart physical effort. (B) Schematic illustration of the task. The reward volume is indicated by the number of drop symbols, and the climbing effort by the black trace of feeder platform height. The sequence of trial blocks (background shading) is repeated after drug injection. (C) Representative sample of a histological section showing the endpoints of electrodes (arrows). The inset show the relative position of the section aligned to a standard rat brain atlas (Paxinos and Watson, 2014). AMPH increases running speed and decreases reward consumption time AMPH administration in our animals evoked typical locomotor effects in a dose-dependent manner. It increased the median running speed (Figure 2A; Kruskal-Wallis, χ2(3)>109; p<0.0001). This effect peaked at 1.0 mg/kg because rats became more frequently disengaged in the task at higher doses. Indeed, the amount of off-task behavior (circling, pausing, backtracking) increased with dose (Figure 2B; main effect ANOVA, F3,18 = 3.36; p=0.042; power = 0.76). Off-task behaviors became so prominent at 2.0 mg/kg that animals did not complete a sufficient number of trials for analysis. Data at this concentration is therefore excluded from the present report. Figure 2 with 1 supplement see all Download asset Open asset Task performance after AMPH injection. (A) Change of median running velocity after injection of AMPH with respect to values before injection. (B) Mean change in the relative proportion of trials with off-task behavior. (C) Change in the median time of occupancy at the start feeder. (D) Representative examples of running path superimposed for all trials in one session before (blue) or after (red) injection of vehicle or AMPH. (E) Mean change in running path roughness of all rats for the task epoch from the target feeders back to the start feeder. (F) Change in roughness of running path from the start feeder to target feeders, as measured by mean change in Hausdorff fractal dimension for all rats. Error bars show the Standard Error of the Mean (SEM) or Median (SEMd). Here and following figures: asterisks (*) indicate statistically significant differences at α = 0.05 with Bonferroni correction; only comparisons with saline are illustrated. AMPH administration also produced typical effects on reward-related behaviors. The rats’ occupancy time at the reward feeders decreased with increasing dose (Kruskal-Wallis median test, χ2(3)>136, p<0.0001 for all feeders). This occurred at the start feeder (Figure 2C) and target feeders (Figure 2—figure supplement 1). Such decreased feeder engagement after AMPH is consistent with previous reports (Randrup et al., 1963; Randrup and Munkvad, 1967; Blundell et al., 1976; Blundell et al., 1979; Leibowitz et al., 1986; Floresco et al., 2008b). In sum, moderate doses of AMPH in this task maintains animals’ engagement in the task, even though they appear less motivated to consume the reward. We next investigated effects of AMPH on running path. The reasons are twofold. First, ACC activity is highly sensitive to the running path, so any gross changes in path trajectory or variance may confound the decoding of other information (Euston and McNaughton, 2006). Second, it provides an additional indicator of task engagement. We therefore use the distribution of running path smoothness as an additional measure of task engagement and/or psychomotor effects. When analyzed over the entire track, the running path of rats became more variable after AMPH, and this effect increased as the dose of AMPH increased (Figure 2E; ANOVA, F3,18 = 3.47; p=0.038; power = 0.77). This is particularly evident on the return arms of the track from the target feeders back to the starting feeder. The inter-trial path variance was not significantly affected in the segment from the starting feeder to target feeders (Figure 2F; ANOVA, F3,18 = 1.83; p=0.178), although it trended to be higher at 1.5 mg/kg. These data indicate that task performance is not disrupted by lower doses, but begins to deteriorate at 1.5 mg/kg. The behavioral data suggest that AMPH’s effect in this study is typical of past studies, in which lower doses facilitate task engagement, but high doses disrupt it. Furthermore, measures of reward consumption decrease with increasing AMPH dose. ACC signals effort and reward The ACC is well known to signal effort and reward. In order to visualize where in the present task ACC neurons signaled these variables, we segmented the track into 36 spatial bins and tested if the mean firing rate of each cell in each bin was significantly different during high-effort versus low-effort trials (t-test, p<0.05). Nearly 20% of cells signaled upcoming effort while on the approach to the barrier, and this proportion ramped up to nearly 35% during the climb/jump to the reward platform (Figure 3A, solid line). Conversely, about 10–15% of ACC cells in our sample encoded reward, and this proportion did not vary much during the task (Figure 3B, solid line). This is consistent with previous reports (Cowen et al., 2012). Administration of vehicle (saline) did not grossly affect the relative proportions of effort-related or reward-related cells (Figure 3A–B, dotted lines). This suggests that the relative proportion of cells signaling these task variables is consistent throughout the session. This is important because we use a within-session task design in order to observe AMPH-related changes in signaling of the same set of neurons, and some neural correlates may change during the session. Indeed, we found that the mean firing rate is lower after vehicle administration (Figure 3C), which almost certainly reflects the typical decrease in firing rate as sessions progress. This decrease is attenuated as AMPH dose increases, and the firing rate is increased at the highest dose relative to the first half of the session (Figure 3C; F3,1262 = 7.05; p=0.0001). This increase of ACC firing rate by AMPH is consistent with previous reports (Lapish et al., 2015), and again suggests that AMPH is having a typical effect in the present study. Figure 3 Download asset Open asset Proportion of ACC neurons encoding effort or reward. (A) Mean proportion of recorded neurons that discriminate barrier height in pre- or post-injection conditions. The proportion increases up to the barrier traverse. The shaded region surrounding the curves indicates SEM. Background colors correspond to each of the 36 spatial bins of the maze shown in the inset. (B) Mean proportion of recorded neurons responding to the reward. (C) Relative change in mean firing rate after injection (mean and SEM). AMPH compresses the encoding of utility by single-units The analysis of costs and benefits is typically formalized through the concept of utility, which can include many features (Glimcher et al., 2009). Here, we consider the joint encoding of effort and reward by individual ACC neurons. We use a linear regression approach, which is a standard method for discriminating when a continuous or binary predictor variable (reward or effort here) is informative of another variable (firing rate). We first computed the correlation of each neuron’s firing rate with effort, and independently computed its correlation with reward volume, in each spatial bin from the starting feeder to the target feeders. We then analyzed the distribution of all cells in the effort-reward space for each spatial bin using principle component analysis (PCA). The first principle component (PC) reveals the primary axis of variance. The distribution of points in the effort-reward space prior to drug reveals two interesting features (Figure 4). First, the distribution of points has a downward diagonal trend (i.e. the slope of the first PC is negative). This indicates that the joint encoding of reward and effort by individual ACC neurons is typically anti-correlated, which is expected of a utility signal. Further, the maximum variance of data is explained by neurons located in both the second (QII) and fourth (QIV) quadrants (Quadrants are numbered as in geometry, starting from the upper right corner and progressing in an anticlockwise direction). Neurons of quadrants QII and QIV have opposing signaling of utility. Neurons of QIV tend to generate more action potentials for high-utility conditions (i.e. when the effort is low or the reward volume is large), and fire less in low-utility conditions (i.e. high-effort or small-reward). Neurons in QII exhibit the inverse relationship among firing and value. The second revealing feature is that ACC cells are correlated with future reward and effort contingencies. This is evident by the non-uniform distribution of points in the spatial bin immediately following departure from the start feeder (Figure 4A, explained variance is 63, 62, 61, 61% for pre-inj, and 62, 63, 59, 57% for post-inj with increasing dose). This is nearly 100 cm in advance of the barrier and target feeder that are the basis of the correlation. Figure 4 with 1 supplement see all Download asset Open asset Effect of AMPH on the encoding of reward and effort by single neurons. (A) Joint effort-reward encoding by ACC neurons in all sessions in the spatial bin immediately following departure from the starting feeder (star in inset). Each black circle shows the Pearson correlation coefficient between the amount of effort and firing rate of one neuron, plotted against the correlation value for reward volume and firing rate of the same neuron. The lines indicate the first principal component (PC) coefficient during pre-injection (blue) and post-injection (red) for all units in the dataset recorded for each dose. (B) First PC of the effort vs reward correlation distribution for cells recorded in all sessions with similar treatment. Each line represents one spatial bin from center feeder to side feeder. Blue lines indicate PCs prior to injection, and red lines indicate the PC for the same cells after the injection. Statistical difference of the coefficient distributions pre versus post injection is computed by the Kuiper two-sample test, and indicated by asterisks: * significant at p<0.05, **: p<0.01, ***: p<0.001. (C) AMPH-evoked changes in mean explained variance by the first PC of effort-reward distributions, showing that the relationship between reward and effort for individual cells breaks down under higher AMPH doses. (D) Post-injection change of the proportion of effort-selective cells, indicating that effort signaling is not reduced by AMPH. (E) Post-injection change in the proportion of reward-selective cells, showing that AMPH decreases reward encoding as measured by the number of cells discriminating reward volume. (F) Post-injection change in the absolute mean difference of firing rate of effort discriminating cells among trials of large vs small reward. The signaling of reward volume is decreased under 1.0 and 1.5 mg/kg. (G) AMPH-related change in the variance of neural firing around the mean firing rate in large and small reward (i.e. class-conditioned variance). The variance is not statistically different under AMPH, but tends to decrease under 1.0 mg/kg and increase under 1.5 mg/kg. Data show mean and SEM. Asterisks (*) indicate statistically significant differences at p<0.05; only comparisons with vehicle are illustrated. The slope of the first PC becomes steeper with increasing AMPH (Figure 4A-B), indicating a loss of neural firing correlation with reward. This relationship holds for the remainder of the spatial bins (Figure 4—figure supplement 1). Circular statistical analysis of the PCs from each spatial bin (first pooled over all sessions with similar treatment) reveals that the post-injection PCs are significantly more vertical than the pre-drug condition for AMPH, whereas saline injection has no effect (Figure 4B; Kuiper two-sample test, k = 160; p=0.028 for 0.5 mg/kg, k = 176; p=0.007 for 1 mg/kg, k = 240; p<0.001 for 1.5 mg/kg AMPH). To ensure that this is not an effect of heterogeneous sample sizes among conditions, we ran a bootstrap analysis (100 repetitions) in which we randomly sub-sampled the data to obtain the same number of neurons for each condition. The PCs were highly stable, and the results did not change with down sampling. Moreover, the percentage of total variance explained by the first PCs remains above 60% for all conditions, which further indicates that the results of the PCA are reliable. The rotation of neural tuning toward the vertical effort axis indicates that the encoding of reward is ‘compressed’ more than is the encoding of effort. Furthermore, the explained variance by the first principal component is significantly decreased at 1.0 and 1.5 mg/kg, which further indicates an overall loss of utility signaling at higher doses of AMPH (Figure 4C.; ANOVA F3,60 = 17.3454; p=3 x 10−8). This effect might be through a breakdown in the encoding of effort or reward (e.g. points moving toward the origin), or an increased dispersion in effort-reward coding by shifting some neurons to the first quadrant in which the neurons respond positively to both effort and reward. In either case, these neurons fail to encode utility. Note that this analysis includes all neurons so as to eliminate any selection bias that may occur by first categorizing cells as utility or other classes based on statistical thresholds. We next analyze statistics of cells first identified as encoding reward or effort. The preceding analysis suggests that neurons jointly encoding both effort and reward tend to lose sensitivity to reward more so than effort. We next sought to determine changes in the signaling of these variables independently, by computing the proportion of cells with significantly different firing rate for different levels of effort, or reward (ANOVA; p<0.05, and moderatley large effect size of partial η2 >0.138). The proportion of units encoding effort shows a weak ‘inverted U’ shape with increasing AMPH, driven by the increase in the proportion of effort cells at 0.5 mg/kg (Figure 4D; ANOVA, F3,60 = 3.73; p=0.016). No dose of AMPH caused a significantly different proportion than vehicle. Conversely, the proportion of units encoding reward strongly decreases with increasing AMPH (Figure 4E; ANOVA, F3,60 = 46.04; p=14 x 10−16). In sum, high doses of AMPH reduce reward signaling in ACC, but do not appear to reduce signaling of effort. This is consistent with our behavioral observations that AMPH-treated rats appear less interested in consuming the reward. These results therefore suggest that reward is devalued by AMPH to a greater extent than effort. The reduced number of reward signaling cells following AMPH could occur because the mean difference of firing rate in the two conditions diminishes (i.e. attenuated signal), or because the variation of firing increases (i.e. increased noise). We therefore computed signal amplitude and variation before and after AMPH, in all cells significantly discriminating reward value prior to drug. The difference of firing rates for large and small rewards decreased as AMPH increased (Figure 4F; F3,1935 = 4.1; p=0.006). Note that the difference is negative after vehicle, likely reflecting reduced motivation for the reward as the animal becomes sated. AMPH appears to accelerate this process. Conversely, the standard deviation of firing from trial to trial was not significantly different under AMPH (Figure 4G; F3,1935 = 2.06; p=0.104). Nonetheless, it shows a trend in which variability of firing is reduced by 1.0 mg/kg, but increased by 1.5 mg/kg. This is consistent with changes in ensemble variance shown later in this report. Note that the standard deviation is computed independently for small and large rewards before averaging, so the reduction is not a consequence of reduced difference of means among large and small reward trials. In sum, these data indicate that the loss of reward signaling for intermediate doses of AMPH (0.5 and 1.0 mg/kg) is primarily due to the signal reduction, rather than an increase in noise. At higher doses, both may play a role. Ensemble ACC activity encodes task epoch, task features, and past/present/future events We next conducted a state-space analysis of simultaneously recorded neurons in order to assess how AMPH affects temporally evolving patterns of neural ensembles. We used a method termed Gaussian Process Factor Analysis (GPFA) to reduce the dimensionality of the data. This algorithm is particularly advantageous for producing smooth trajectories in low-dimensional space from high-dimensional processes with discrete events, such as action potentials (Yu et al., 2009). Similar to PCA, GPFA serves to capture as much variance as possible and does not optimize for any particular information present in the data (e.g. reward, location). In the 3D space of the first three GPFA factors, our ACC data form trajectories that move smoothly in the reduced space as the trial progresses (Figure 5). The trajectories discriminate task epochs, but are highly similar across trials of the same type. The trajectories diverge for trials of different effort or reward (Figure 5, right panel). For instance, the trajectory at the barrier climb deviates in high-effort trials as compared to low-effort ones. This is expected from the single-unit analysis, which showed that approximately 1/3 of ACC neurons discriminated effort during this epoch. Note that the ACC trajectory diverges widely on the approach to the turn (red shading), which is well before the barrier climb. This is consistent with the single unit analysis showing correlation of units on the middle segment, and further suggests that the ACC is sensitive to upcoming events. Figure 5 Download asset Open asset Neural trajectories encode task sequence. Plots show trajectories of neural encoding projected into low-dimensional space, and are color-coded by the sequential task epochs shown in the inset. Several trials are superimposed, revealing that the patterns of neural activity are distinct for each epoch but are highly similar across trials of the same type. The trajectories diverge for trials requiring different effort (right panel), and the divergence begins well before the actual barrier traverse, which occurs in the green region, as seen by divergence of encoding in the central stem of the track (red region). The state-space analysis reveals another interesting novel phenomenon that cannot be discriminated directly from individual units. We analyzed trials in which the animal approached the barrier, then turned around and backtracked to the center feeder, and finally completed the trial by approaching and climbing the barrier. During this second approach and trial completion, the ACC diverged greatly from its typical trajectory during this task epoch, even though the path and velocity of the rat was very similar during both approaches (Figure 6). It thus appears that the ACC is sensitive to the recent history (or context) in which the task-related activity occurs. In sum, these data indicate that the trajectory of ACC activity in the reduced space is highly sensitive to the task epoch, task features (e.g. barrier climb), as well as the sequence of past and future events. Moreover, it shows that deviations from the trajectory are concomitant with off-task behavior. Changes in the trajectories, therefore, likely reveal changes in the neural processes directing task engagement and execution. Figure 6 Download asset Open asset Ensemble encoding is task-specific. Top plot shows ACC trajectories of population activity in low-dimensional space for several trials in one session, including three trials with off-task behavior (identified by trial numbers). In these cases, the animal approached the barrier, then backtracked from the barrier to the start feeder before returning to commit the climb and reach the target feeder. The running paths and velocity vectors are shown in the lower panels. Note that the neural trajectories deviate largely when the animal is on the second approach and climb (dashed lines in upper panel) as compared to trials with no off-task behavior (solid red-green-orange sequence of lines). This deviation occurs, even though the paths and velocity of the initial approach (fuchsia) and re-approach (black) are similar. This suggests that ACC is sensitive to past events – in this case, off-task behavior. AMPH contracts ACC state-space occupancy at low doses, and expands it at high doses We next examined the effect of AMPH on neural trajectories. Low-dose AMPH caused a contraction of trajectories in state space for both large-reward trials (Figure 7B; ANOVA, F3,371 = 28.28, p=2 x 10−16) and small-reward trials (Figure 7C; ANOVA, F3,325 = 19.78, p=8 x 10−12). This is not likely explained by differences in running paths. This is because reduced state-space volumes suggest less variance of neural firing. This would require less variance in running paths, which does not occur (Figure 2D-F). It is also possible the first three latent factors capture less variance of the neural activity after AMPH, in which case the reduction in volume is likely an artifact of the methodology. We therefore compared the explained variance of the first three factors and found that it was not reduced by administration of 0.5–1.0 mg/kg AMPH (Figure 7D; confidence intervals do not deviate from 0). The contraction therefore appears to be a neural phenomenon, rather than a methodological artefact. As opposed to the reduction of state-space volume under low-dose AMPH, we instead found that 1.5 mg/kg caused an expansion (Figure 7B-C). Because the roughness of the running path increased at this dose, it is possible that these phenomena are related. Figure 7 Download asset Open asset Dose-dependent contraction of neural trajectories. (A) Example of ACC neural trajectories before injection (left panel), and after injection of 0.5 mg/kg AMPH (right panel) plotted into the same low-dimensional space for one session. The volume enclosed by the envelope (outer hull) of the trajectories contracts after AMPH. (B) Post-AMPH changes in mean trajectory volume over all sessions and trials with large reward. The volume decreases for intermediate doses, and expands at the highest dose. (C) The same dose-dependent volume changes occur for small-reward trials. (D) Changes in the amount of variance explained by the first three factors after AMPH. Medians and 95% confidence intervals are shown, revealing that explained variance is not significantly reduced after AMPH because the confidence intervals encompass 0. (E) Change in SEM of neural trajectories after AMPH on large reward trials averaged over all the first three factors and all spatial bins, showing a reduction in trial-by-trial variance for low-dose AMPH. (F) Same as panel E for small reward trials. (G) Change in SEM for the first factor across task epochs. Data are averaged over all sessions with similar treatment for large reward trials. Asterisks (*) indicate statistically significant differences at p<0.05; only comparisons with vehicle are illustrated. Panels B, C, E, and F show means and SEM. It has been suggested that the neu

Psychostimulants such as d-amphetamine (AMPH) often have behavioral effects that appear paradoxical within the framework of optimal choice theory. AMPH typically increases task engagement and the effort animals exert for reward, despite decreasing reward valuation. We investigated neural correlates of this phenomenon in the anterior cingulate cortex (ACC), a brain structure implicated in signaling cost-benefit utility. AMPH decreased signaling of reward, but not effort, in the ACC of freely-moving rats. Ensembles of simultaneously recorded neurons generated task-specific trajectories of neural activity encoding past, present, and future events. Low-dose AMPH contracted these trajectories and reduced their variance, whereas high-dose AMPH expanded both. We propose that under low-dose AMPH, increased network stability balances moderately increased excitability, which promotes accelerated unfolding of a neural ‘script’ for task execution, despite reduced reward valuation. Noise from excessive excitability at high doses overcomes stability enhancement to drive frequent deviation from the script, impairing task execution.

Communication during play fighting can be crucial in facilitating and sustaining contact. As juvenile rats play mostly in the dark, visual signals are unlikely to fulfill this function. However, during play, rats have a variety of ultrasonic calls that are emitted and there is growing evidence that some of these calls may provide a means of communication. Particular calls are associated with particular actions and in the present paper we show that specific calls by one rat changes the probability of the actions taken by another. We have found that some calls appear to act as a means of sustaining the animals' playful mood and so facilitate the occurrence of play, some calls ensure that the interaction remains playful and so avoid escalation to aggression, and some appear to promote reciprocal exchanges and so ensure that playful contact continues.

Neocortical sleep spindles have been shown to occur more frequently following a memory task, suggesting that a method to increase spindle activity could improve memory processing. Stimulation of the neocortex can elicit a slow oscillation (SO) and a spindle, but the feasibility of this method to boost SO and spindles over time has not been tested. In rats with implanted neocortical electrodes, stimulation during slow wave sleep significantly increased SO and spindle rates compared to control rest periods before and after the stimulation session. Coordination between hippocampal sharp-wave ripples and spindles also increased. These effects were reproducible across five consecutive days of testing, demonstrating the viability of this method to increase SO and spindles.

Interaction between hippocampal sharp-wave ripples (SWRs) and UP states, possibly by coordinated reactivation of memory traces, is conjectured to play an important role in memory consolidation. Recently, it was reported that SWRs were differentiated into multiple subtypes. However, whether cortical UP states can also be classified into subtypes is not known. Here, we analysed neural ensemble activity from the medial prefrontal cortex from rats trained to run a spatial sequence-memory task. Application of the hidden Markov model (HMM) with three states to epochs of UP–DOWN oscillations identified DOWN states and two subtypes of UP state (UP-1 and UP-2). The two UP subtypes were distinguished by differences in duration, with UP-1 having a longer duration than UP-2, as well as differences in the speed of population vector (PV) decorrelation, with UP-1 decorrelating more slowly than UP-2. Reactivation of recent memory sequences predominantly occurred in UP-2. Short-duration reactivating UP states were dominated by UP-2 whereas long-duration ones exhibit transitions from UP-1 to UP-2. Thus, recent memory reactivation, if it occurred within long-duration UP states, typically was preceded by a period of slow PV evolution not related to recent experience, and which we speculate may be related to previously encoded information. If that is the case, then the transition from UP-1 to UP-2 subtypes may help gradual integration of recent experience with pre-existing cortical memories by interleaving the two in the same UP state. This article is part of the Theo Murphy meeting issue ‘Memory reactivation: replaying events past, present and future'.

Neurons which fire in a fixed temporal pattern (i.e., “cell assemblies”) are hypothesized to be a fundamental unit of neural information processing. Several methods are available for the detection of cell assemblies without a time structure. However, the systematic detection of cell assemblies with time structure has been challenging, especially in large datasets, due to the lack of efficient methods for handling the time structure. Here, we show a method to detect a variety of cell-assembly activity patterns, recurring in noisy neural population activities at multiple timescales. The key innovation is the use of a computer science method to comparing strings (“edit similarity”), to group spikes into assemblies. We validated the method using artificial data and experimental data, which were previously recorded from the hippocampus of male Long-Evans rats and the prefrontal cortex of male Brown Norway/Fisher hybrid rats. From the hippocampus, we could simultaneously extract place-cell sequences occurring on different timescales during navigation and awake replay. From the prefrontal cortex, we could discover multiple spike sequences of neurons encoding different segments of a goal-directed task. Unlike conventional event-driven statistical approaches, our method detects cell assemblies without creating event-locked averages. Thus, the method offers a novel analytical tool for deciphering the neural code during arbitrary behavioral and mental processes.

Similar to drugs of abuse, random-ratio reward schedules are highly motivating and, in humans, are thought to foster gambling addiction. Animal gambling models, however, have not yet demonstrated the compulsivity so characteristic of drug addiction. Three criteria have been used to evaluate addiction-like behavior in drug models: (1) response inhibition when reward is not available, (2) persistence under a progressive ratio schedule, in which the response-to-reward ratio is stretched, and (3) persistence in spite of punishment. We tested whether prolonged exposure (6 weeks) to a gambling-like reward schedule would induce addiction-like symptoms in rats. In two studies, separate groups were trained to respond to either random- or fixed-ratio schedules for food reward. We found that rats trained on random-ratio schedules showed higher response rates and dramatically shorter pauses after rewards. Tests of addiction-like behavior, however, were largely negative. Response rates were not different during cued no-reward periods nor when reward was coupled with punishment. We also found no group differences when food was devalued nor in reinstatement of reward-seeking after a 1-week delay. The sole exception to this pattern was that rats in the second experiment showed greater persistence on a progressive ratio test. After experiment two, subjects were also orally administered pramipexole, which caused increased perseveration during progressive ratio testing, especially in the random ratio group. While, it is possible that longer training or more appetitive rewards might have led to addiction-like behavior, our results, on the surface, suggest that random-ratio schedules are motivating but not addictive.

Play fighting in juvenile rats is associated with a high occurrence of 50 kHz vocalizations. These calls are varied in form, ranging from long, flat calls to short, frequency-modulated ones. We hypothesize that at least some types of calls serve as play signals to facilitate play. In the present study, pairs of juvenile male rats that were unfamiliar with one another were paired in a neutral test enclosure to which they had been habituated. Video and audio records were made of the encounters. Pairs were of two types: both pairs could vocalize or only one partner could do so. There were some differences between the play of pairs containing a devocalized partner, but overall, the pattern of play, the frequency and types of calls were similar between the two types of pairs. We used a Monte Carlo shuffling technique to analyse the correlations between the playful actions performed and the types and frequencies of various 50 kHz calls that were produced. The analyses revealed that there were strong associations between types of calls and types of social contact: an approach followed by playful nape contact was associated with calls, but an approach followed by nonplayful contact (e.g. anogenital sniffing) was not. Similarly, different calls were associated with different actions, such as nape contact, evade and wrestling, with most of these calls being uttered by the initiator of the action, not the recipient. However, coordinating calls reciprocally with complementary calls uttered by participants as they engaged in complementary actions (e.g. attacking, being attacked) appeared to be a way in which calls could potentially be used as play signals to influence the ongoing cooperation needed to sustain play fights. (C) 2018 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Ultrasonic vocalizations of around 50 kHz are emitted extensively during the play fighting of rats, reflecting a positive affective state. Some of these calls are associated with facilitating play—at an emotional level they appear to promote the playful mood of both the emitter and the recipient and at a tactical level they appear to coordinate playful contact and promote cooperative actions. The tactical role of these calls in averting escalation to serious fighting becomes more critical when adult rats engage in play. The experience of play fighting in the juvenile period promotes the development of social skills and does so, in part, by increasing the role of the prefrontal cortex in regulating social actions. There is emerging evidence that the experience of the diverse communicatory uses of 50-kHz calls during juvenile play may similarly facilitate the development of nuanced communicatory skills.

The anterior cingulate cortex (ACC) encodes information supporting mnemonic and cognitive processes. We show here that a rat’s position can be decoded with high spatiotemporal resolution from ACC activity. ACC neurons encoded the current state of the animal and task, except for brief excursions that sometimes occurred at target feeders. During excursions, the decoded position became more similar to a remote target feeder than the rat’s physical position. Excursions recruited activation of neurons encoding choice and reward, and the likelihood of excursions at a feeder was inversely correlated with feeder preference. These data suggest that the excursion phenomenon was related to evaluating real or fictive choice outcomes, particularly after disfavoured reinforcements. We propose that the multiplexing of position with choice-related information forms a mental model isomorphic with the task space, which can be mentally navigated via excursions to recall multimodal information about the utility of remote locations.

Cell assembly is a hypothetical functional unit of information processing in the brain. While technologies for recording large-scale neural activity have been advanced, mathematical methods to analyze sequential activity patterns of cell-assembly are severely limited. Here, we propose a method to extract cell-assembly sequences repeated at multiple time scales and various precisions from irregular neural population activity. The key technology is to combine “edit similarity” in computer science with machine-learning clustering algorithms, where the former defines a “distance” between two strings as the minimal number of operations required to transform one string to the other. Our method requires no external references for pattern detection, and is tolerant of spike timing jitters and length irregularity in assembly sequences. These virtues enabled simultaneous automatic detections of hippocampal place-cell sequences during locomotion and their time-compressed replays during resting states. Furthermore, our method revealed previously undetected cell-assembly structure in the rat prefrontal cortex during goal-directed behavior. Thus, our method expands the horizon of cell-assembly analysis.

Rat ultrasonic vocalizations have been suggested to be either a byproduct of physical movement or, in the case of 50-kHz calls, a means to communicate positive affect. Yet there are up to 14 distinct types of 50-kHz calls, raising issues for both explanations. To discriminate between these theories and address the purpose for the numerous 50-kHz call types, we studied single juvenile rats that were waiting to play with a partner, a situation associated with a high number of 50-kHz calls. We used a Monte-Carlo shuffling procedure to identify vocalization-behavior correlations that were statistically different from chance. We found that certain call types ("split", "composite" and "multi-step") were strongly associated with running and jumping while other call types (those involving "trills") were more common during slower movements. Further, non-locomotor states such as resting and rearing were strongly predictive of a lack of vocalizations. We also found that the various sub-types of USVs can be clustered into 3-4 categories based on similarities in the way they are used. We did not find a one-to-one relationship between any movements and specific vocalizations, casting doubt on the motion byproduct theory. On the other hand, the use of specific calls during specific behaviors is problematic for the affect communication hypothesis. Based on our results, we suggest that ultrasonic calls may serve to coordinate moment-to-moment social interactions.