A dopaminergic reward prediction error signal shapes maternal behavior in mice

Yunyao Xie; Longwen Huang; Alberto Corona; Alexa H Pagliaro; Stephen D Shea

doi:10.1016/j.neuron.2022.11.019

. Author manuscript; available in PMC: 2024 Feb 15.

Published in final edited form as: Neuron. 2022 Dec 20;111(4):557–570.e7. doi: 10.1016/j.neuron.2022.11.019

A dopaminergic reward prediction error signal shapes maternal behavior in mice

Yunyao Xie ^1,², Longwen Huang ^1,², Alberto Corona ¹, Alexa H Pagliaro ¹, Stephen D Shea ^1,^3,^*

PMCID: PMC9957971 NIHMSID: NIHMS1856976 PMID: 36543170

SUMMARY

How social contact is perceived as rewarding and subsequently modifies interactions is unclear. Dopamine (DA) from the ventral tegmental area (VTA) regulates sociality, but the ongoing, unstructured nature of free behavior makes it difficult to ascertain how. Here we tracked the emergence of a repetitive stereotyped parental ‘retrieval’ behavior, and conclude that VTA DA neurons incrementally refine it by reinforcement learning (RL). Trial-by-trial performance was correlated with the history of DA neuron activity, but DA signals were inconsistent with VTA directly influencing the current trial. We manipulated the subject’s expectation of imminent pup contact and show that DA signals convey reward prediction error, a fundamental component of RL. Finally, closed-loop optogenetic inactivation of DA neurons at the onset of pup contact dramatically slowed emergence of parental care. We conclude that this component of maternal behavior is shaped by an RL mechanism in which social contact itself is the primary reward.

eTOC blurb

The mechanism by which social contact is perceived as rewarding and shapes future interactions is poorly understood. Xie et al. find that midbrain dopamine neurons signal a social reward prediction error, instructing the emergence of maternal behavior by reinforcement learning, and affecting future, but not current, trials.

INTRODUCTION

For many animals, including humans, social contact is highly rewarding ¹. Nevertheless, how social rewards are encoded in the brain and used to modify social behavior is poorly understood. Recent studies revealed some of the neural substrates for regulating sociality and sensing social reward ^2–7, but the temporally unstructured nature of most social interactions presents a challenge for dissociating the immediate motivational and motor influences of DA from the cumulative reinforcing influence of interactions on behavior. Here we use the gradual emergence and refinement of maternal care in virgin female mice to examine the relationship between midbrain DA reward signals and performance over many iterations of a stereotyped ethological social behavior.

A confluence of data and theory supports a framework for understanding how organisms adapt their behavior to maximize reward known as ‘reinforcement learning’ (RL) ^8,9. Two crucial features of most RL models are iteration and reward prediction. Over many trials, an agent selects actions predicted to bring reward, and on each trial it compares the received reward with the predicted reward. The difference, called ‘reward prediction error’ (RPE), is then used to update subsequent actions and reward predictions. There is a wealth of evidence that RPE is explicitly signaled by midbrain DA neurons in response to rewards ^10,11, however DA has other well-established roles in goal-directed behavior, including motivational drive ^12,13 and motor control and invigoration ^14–17. We took advantage of the reliable and repetitive nature of maternal pup retrieval to distinguish between these potential contributions of the midbrain DA system to its emergence.

When first exposed to offspring, over several days, primiparous and virgin female mice exhibit increasingly rapid and reliable retrieval of pups that become separated from the nest ^18,19. We chose to examine the role of midbrain dopamine signals in the emergence of maternal retrieval for several reasons. First, access to pups positively reinforces behavior ^20–23 with comparable efficacy to cocaine ²⁴. Second, the dopaminergic system has been linked to the establishment and motivation of pup retrieval ^25–31. For example, maternal experience is associated with reduced baseline DA activity and enhanced DA activity in response to pup interactions in the mesolimbic pathway ^32–39. Third, the repetitive, stereotyped nature of retrieval facilitated a trial-by-trial analysis of DA signals and performance.

We measured fluctuations in mesolimbic DA neuron activity during pup retrieval as the behavior emerged in virgin female mice. We took advantage of the iterative structure of pup retrieval to track changes in DA neuron signaling over many trials. Figure 1a shows three distinct patterns of results we expected if VTA DA signals predominantly act to motivate initiation of retrieval (model 1), invigorate motor activity during retrieval (model 2), or signal RPE with respect to pups (model 3). We find that DA signals during pup retrieval are consistent with an RPE used to reinforce future performance, rather than direct motivation or motor invigoration of the current trial. Consistent with this model, when we optogenetically inactivated VTA DA neurons as the female contacted the pup, the female did not improve. However, inactivating only on alternating trials revealed that the acquisition of retrieval behavior was still slowed without directly impairing performance acutely. We conclude that VTA DA shapes this form of maternal care by the cumulative reinforcing effect of an RPE signal. These findings are most consistent with the incremental improvement in the efficiency of retrieval resulting from the female learning the value of retrieving pups.

Figure 1: — (a) Schematic depicting three possible models for the trial dynamics of the mouse’s velocity and the magnitude of optically measured activity in VTA DA neurons: (1) VTA DA neurons motivate initiation of pup retrieval by firing primarily at the onset of approach. (2) VTA DA neurons invigorate motor activity by firing during retrieval. (3) VTA DA neurons signal reward prediction error, therefore they fire primarily at pup contact. In models 1 and 2, VTA DA neuron signals tend to increase as learning progresses. In model 3, the signals are expected to decrease as learning progresses.

(b) Intersectional viral expression strategy for fiber photometry.

(c) Co-localization of GCaMP7f and VTA DA neurons. Scale bar=100 µm.(d) Raw trace showing ΔF/F of activity in VTA DA neurons from a nulliparous female mouse on P0. Red arrows indicate when the female lifted the pup in its mouth.

(e) Heatmaps depicting Z-scored ΔF/F (bottom panel) and mouse velocity (top panel) for 14 trials of pup retrieval in one daily session. Data are from the same mouse and session as in (d).

(f) Plot of mean ΔF/F (black) for all trials in (e) aligned to mean velocity (red) for all trials. The gray shaded region around the ΔF/F trace indicates ± SEM.

See also Movie S1.

RESULTS

To initially observe DA neuron activity during pup retrieval, we injected the ventral tegmental area (VTA) of virgin female DAT-Cre^+/− mice with Cre-dependent adeno-associated virus (AAV) driving expression of jGCaMP7f ⁴⁰, a genetically-encoded Ca²⁺ sensor (Figure 1b). This resulted in expression of GCaMP in VTA DA neurons (Figure 1c) with high sensitivity and specificity: 91.6 ± 1.9% tyrosine hydroxylase (TH+) neurons were labeled with GCaMP and 98.1 ± 0.8% GCaMP+ neurons were labeled with TH. We then used fluorescence fiber photometry to measure fluctuations in population neural activity during maternal interactions.

When nulliparous female subjects were placed into their home cage with familiar pups, individual retrieval events were each accompanied by a brief (2 – 3 s) increase in fluorescence (Figure 1d). These peaks were generally consistent across all events within a day in terms of amplitude and temporal relationship to the retrieval time as measured by the automatic tracking of the mouse (DeepLabCut ⁴¹) (Figure 1e–f). The timing and shape of the retrieval events were very similar to measurements of DA release in the nucleus accumbens core ³⁹. This suggests that despite the well-known heterogeneity of VTA neurons, we are still able to detect signals that are relevant to reward and action value with respect to retrieval.

In the same recording sessions, subjects exhibited pup-directed grooming and nesting behaviors, accompanied by a slight increase in the responses of DA neurons at the onset (Figure S1a and S1b). However, these signals were dramatically smaller than retrieval responses, strongly suggesting that motor aspects of pup contact did not explain the large transients evoked during retrieval.

VTA DA neuron signals during retrieval are inversely correlated with performance

As expected, over days of practice retrieving pups, nulliparous females tended to improve the speed and efficiency of their pup retrieval (Figure 2e). We therefore examined the relationship between the GCaMP fluorescence and behavioral performance over many individual retrieval events in mice subjected to three different types of maternal experience (Figure 2b). Figure 2a shows the predicted results for each of the three models in Figure 1a (motivation, motor, and RPE) when comparing retrieval performance and mean VTA signal amplitude by day. Note that if VTA DA neurons are encoding an acute motivation to initiate retrieval (blue) or direct invigoration of motor behavior (purple), we expect to see a positive relationship between retrieval activity and retrieval performance. In contrast, as the mouse improves, reward becomes expected, so if VTA DA neurons are signaling RPE, we predict an inverse relationship for those variables (green).

We performed daily pup retrieval tests (P0 – P5) on virgin mice while continuously co-housed (CH) with a new mother and her pups (Figure 2b–c). We observed transient increases in fluorescence that began at pup contact (Figure S1c) and peaked just after the female lifted the pup (Movie S1). The amplitude of signals aligned to contact was highly correlated with that aligned to retrieval (Figure S1d). Retrieval-evoked VTA activity was initially of comparable magnitude to responses to consumption of an unexpected treat (1 per day at a random time; Figure S2). Signals were largest on P0, and on subsequent days, the amplitude decreased sharply (Figure 2c–d).

To assess whether decreasing DA neuron signals were due to continuous exposure to pups, we tested retrieval in a separate ‘Non co-housed’ (NCH) cohort that only encountered pups during one daily round of testing (5 trials) and then remained in their home cage with other virgins only (Figure 2b). In contrast to CH mice, NCH subjects exhibited large amplitude GCaMP signals through the end of the experiment (P5) (Figure 2c–d).

Finally, we reasoned that the differences in VTA activity and behavioral performance between the CH and NCH groups could directly reflect housing conditions, or simply the additional interaction time experienced by the CH group. We therefore performed retrieval tests on a third cohort we refer to as ‘High experience, non co-housed’ (HE/NCH; Figure 2b). These mice were housed separately from the dam and pups, but they were given many more trials per day (~24). Their DA neuron activity diminished rapidly after P0 like the co-housed group (Figure 2c–d), showing that the decline was predominantly related to the quantity of practice the mice had and not simply an effect of housing condition. While all three groups showed similar improvements in some aspects of pup-related behaviors (Figure S3), the changes in DA signals were related to the strength and pace of learning, as measured by improvement in ‘performance index’ (see Methods). A larger value indicates lower latency to retrieve. The CH and HE/NCH mice showed indistinguishably robust and rapid learning (Figure 2e). The relative signal magnitude in each group over days was unaffected by aligning to contact (Figure S1e), and all groups initially had comparable signals (Figure 1f). In contrast, the persistent, high-amplitude signals in NCH were accompanied by poorer retrieval (Figure 2e). Behavioral improvement across daily sessions in all groups was also evident in the velocity as females approached pups (Figure 3c). Learning in all cohorts was closely related to VTA activity because performance on each day was inversely correlated with response magnitude (Figure 2f–h). Thus, comparing mean VTA signals with retrieval performance at the session level reveals a pattern that is inconsistent with retrieval-related DA activity invigorating or motivating initiation of pup retrieval on the same trial.

Figure 3: — (a) Predicted results of each model for the relationship between trial performance and VTA DA neuron signals for the same trial.

(b) Predicted results of each model for the relationship between session performance and mean VTA DA neuron signals for the previous trial.

(c) Plot of mean ± SEM of approach velocity for all days separated by cohort (two-way ANOVA with Tukey correction, main factor (day): *p < 0.0001*, main factor (mouse type): *p < 0.0001*; # and * represent significant difference in the CH vs the HE/NCH groups, and in the NCH vs the HE/NCH groups respectively, *p < 0.05, **p < 0.01).

(d) Behavioral and neural data from one example mouse. *left*: Aligned vertical plots of the magnitude of VTA DA neuron activity (black trace) and approach velocity (red trace) for each of 140 trials over 6 days. *right*: Heatmap of neural activity for the same 140 trials. Each row is one trial, and rows appear in chronological order from top to bottom.

(e) Scatterplot of the Pearson correlation coefficient (r) for each mouse between the AUC of the ΔF/F trace and (*l-r*) the mean velocity of the mouse returning to the nest, the mean velocity as the mouse approaches the pup, and the difference in approach velocity between the current and the next trial. Filled points show significant correlations (p < 0.05), and the gray points show results from shuffled trials. Mean correlation coefficient did not significantly differ from that of shuffled data for return velocity (n = 16 mice; matched trials: −0.04 ± 0.2; shuffled trials: 0.00 ± 0.0, Wilcoxon matched pairs signed rank test, p = 0.23). Mean correlation coefficient was significantly more negative than that of shuffled data for approach velocity (n = 16 mice; matched trials: −0.43 ± 0.21; shuffled trials: 0.00 ± 0.0, Wilcoxon matched pairs signed rank test, ***p < 0.001). Mean correlation coefficient was significantly more positive than that of shuffled data between the AUC of ΔF/F and the change in approach velocity on the next trial (n = 10 mice; matched trials: 0.20 ± 0.04; shuffled trials: 0.00 ± 0.0, Wilcoxon matched pairs signed rank test, **p < 0.01).

(f) Each column shows the same data sorted differently. In the left panel, the rows of the heatmap are sorted according to increasing return velocity, in the middle panel they are sorted by increasing approach velocity, and in the right panel they are sorted by the change in approach velocity on the next trial. At the bottom, the green traces depict the mean ΔF/F signal for each quintile as denoted by color bar to the left of each of each heatmap.

(g-i) Scatterplots and Pearson correlation of approach velocity versus VTA signal amplitude and change in approach velocity since the previous trial versus the VTA signal amplitude on the previous trial for the CH cohort (g), the NCH cohort (h) and the HE/NCH cohort (i).

Retrieval performance is influenced by the history of DA neuron activity

Our observation that VTA responses to pup contact resembled responses to treats is consistent with past work showing that contact with pups is powerfully intrinsically rewarding ^20,21,24,42. RPE is the difference between expected and received reward, as a reward becomes predictable, RPE decreases. This quantity is commonly signaled by VTA, therefore we speculated that the decrease in VTA responses to pups may reflect an RPE for pup contact. If so, then it should exhibit three properties. First, it should incrementally influence retrieval behavior on a trial-by-trial basis during early maternal experience. Second, as reward becomes increasingly predictable due to a reliable cue, RPE should decrease for the reward and increase for the predictive cue. Third, surprising reward should evoke a larger RPE, while surprising lack of reward should evoke a negative signal.

We tested the first property above by performing a trial-by-trial analysis of the magnitude of VTA activity. We compared the signal with approach velocity (as a measure of trial performance; Figure 3c) on the same trial and the ensuing trial. Figure 3a–b shows the predicted results for each model of the contribution of VTA activity to retrieval: direct motivation and motor drive will increase approach velocity on the current trial and will not affect the subsequent trial. In contrast, RPE will be weaker when performance is high, because the reward is anticipated, and as it reinforces retrieval, it will be correlated to performance on the subsequent trial.

As the virgin mice accrued maternal experience, retrieval time and DA neuron activity steadily decreased. However, both quantities showed considerable trial-to-trial variability (Figure 3d). Therefore, we more closely examined the relationship between behavior and the recent history of VTA responses to pups. First, we compared the mean velocity of the female as she returned to the nest with the pup to the VTA fluorescence from the same trial, and observed a significant correlation in only 1/16 mice (Figure 3e). This lack of a relationship argues against VTA DA neuron activity simply motivating or invigorating retrieval motor behavior. We used approach velocity as a single-trial measure of motivation and performance. This is supported by the observation that like performance index, approach velocity steadily improved over training (Figure 3c). As expected from the analysis of retrieval and VTA at the session level, for all three groups, approach velocity was inversely correlated with the VTA signal on the same trial (Figure 3e). This relationship was seen in 12/16 mice. Interestingly, we found the opposite relationship for adjacent trials. The difference in approach velocity between the current and previous trial was significantly positively correlated with VTA activity on the previous trial (Figure 3e), and in 6/10 individual mice. This was also true when correlating with the absolute approach velocity. All correlations were abolished by shuffling trials (Figure 3e). The positive correlation between DA neuron activity and future retrieval performance extended beyond one trial. Considering only the HE/NCH group, which had a sufficient number of trials/day for this analysis, we found that the improvement in approach velocity was significantly positively correlated with the sum of the DA neuron activity on the previous 2 – 5 trials.

We sorted the ΔF/F traces based on the velocity of the mouse’s return to the nest, the velocity of its approach to the pup, and the change in approach velocity on the subsequent trial, and we plotted the mean VTA DA neuron activity for the trials in each quintile (Figure 3f). Mean ΔF/F traces decrease with increasing approach velocity and increase with increasing difference in approach velocity on the subsequent trial. We also found a significant positive correlation between the change in approach velocity on the subsequent trial and the DA neuron activity to pup retrieval in CH, NCH and HE/NCH mice (Figure 3g–i). Collectively, these correlative results raise the possibility that the mouse updates its behavior trial-to-trial based on the strength of the last VTA DA neuron signal it experienced.

VTA DA neurons signal reward prediction error for pup contact

To test the effect of reward expectation on VTA DA neuron activity, and to observe the dynamics of responses to both pups and a cue signaling an imminent pup interaction, we devised a cued retrieval task (Figure 4a and Methods). Virgin mice began each trial with an empty nest outside a two-chambered structure. A pup was placed in one of two chambers, either an inaccessible chamber or a chamber that could be accessed through a motorized sliding door that was closed between trials. On each trial, one of two tones was played, either a tone (CS+) that usually (>90%) signaled that the pup was in the accessible chamber, or a tone (CS−) that usually (>90%) signaled that the pup was in the inaccessible chamber (CS−). After a fixed delay (1 s from the end of the tone), the sliding door opened and the mouse was permitted to enter the chamber to search for a pup. Figure 4c shows mean VTA signals aligned to the tone, the door opening, and chamber entry, and separated into trials with (black) and without (red) an accessible pup. Trials were randomized within a day and were carried out according to a standardized schedule on P0 – P8 (Figure S4a–b).

VTA DA neuron responses to task events changed over training. Most mice exhibited little response to either tone (Figure 4b). This was surprising because the CS+ predicted a rewarding pup encounter. We did however observe stronger and more consistent responses to the door opening, an event more proximal to pup contact (Figure 4d). On trials where the female encountered and retrieved a pup, we plotted the mean activity aligned to door opening and chamber entry for all mice in 10 trial bins (Figure 4d). Along with strong responses to the door, there were large responses to chamber entry, which were followed quickly by pup contact. Considering only trials where the female encountered a pup, over many trials the entry response grew weaker and the door response grew stronger (Figure 4e). Activity was unchanged over trials without a pup for both door opening and chamber entry (Figure 4e).

Using linear time warping to align the neural data to both the opening of the door and the retrieval of the pup (Figure S4c–f and Methods) revealed precise and distinct responses to the door and retrieval that were significantly different from baseline and each other (Figure 4f–g). Comparing the magnitude of each peak between the first 40 pup trials (‘early’) and the last 40 pup trials (‘late’) showed a significant increase in responses to the door and a significant decrease in responses to pup contact in late trials (Figure 4h). There was no significant change in pre-door opening activity between early trials and late trials (Figure 4i). This pattern of VTA DA neuron activity strongly resembles changes in RPE over time, because RPE decreases for the reward and increases for the predictive cue as reward becomes more predictable.

To examine the effects of expectation on responses to task events, we manipulated the mouse’s prediction of access to a pup. The CS+ and CS− tones were not perfectly reliable cues for whether the pup would be accessible on a given trial. We included up to 10% ‘catch trials’ on which either the CS− tone was played and there was unexpectedly a pup in the chamber (CS− catch trials), or the CS+ tone was played and the pup was unexpectedly absent from the chamber (CS+ catch trials) (Figure 5a; Figure S4a–b). Figure 5b–c compares the mean activity for all mice between catch trials and standard trials. The mean activity upon entry on CS− catch trials was significantly higher than on CS+ trials (Figure 5b–c). In both trials, the mouse experienced the same outcome, but on the catch trial, the presence of the pup was unexpected. In contrast, on CS+ catch trials, the activity just before entry was significantly greater than that on CS− trials, potentially in anticipation of pup contact (Figure 5b–c). Later in the trial, once the mouse had time to search the chamber and found no pup, there was a small, but significant dip in VTA activity (Figure 5b–c). This pattern of results is consistent with VTA signals representing RPE for pup contact, since a surprising reward leads to a larger RPE, while a surprising omission of reward results in a negative RPE value.

Figure 5: — (a) Schematic of cued retrieval task with catch trials.

(b) *top:* Heatmaps depicting the mean activity for each of four trial types (CS+, CS+ catch, CS−, CS− catch). Each row is the mean VTA DA neuron activity for one mouse. *bottom:* Mean ± SEM traces of each trial type comparing entry activity of (*left*) CS− catch trials (red trace) to CS+ trials (black trace) and (right) CS+ catch trials (red trace) to CS− trials (black trace). Gray shaded regions denote the time range for calculating mean signal.

(c) Scatterplot comparing the mean DA neuron activity between normal and catch trials for each of the three windows marked in (a) (n = 12 mice, paired t test comparison with Holm-Bonferroni correction, *p < 0.05, **p < 0.01).

See also Figure S4.

Optogenetic inhibition of VTA DA neuron activity impairs pup retrieval

We tested whether VTA DA neuron activity is required for rapid retrieval learning. DAT-Cre^+/− mice were injected in VTA in both hemispheres with a Cre-dependent AAV driving expression of either the inhibitory optogenetic tool stGtACR⁴³ or EGFP as a control (Figure 6a–b, Figure S5). The large majority of stGtACR and EGFP-expressing cells (95.88% and 97.43% respectively) in the VTA were positive for expression of TH, while 43.81% and 98.54% TH+ neurons also expressed stGtACR and EGFP respectively. Naïve virgin females were tested with 20 daily trials of open field pup retrieval (P0 – P3). On each trial, a barrier was lifted, revealing a pup, and the mouse was free to retrieve it at any time during the trial (up to 300 s) (Figure 6c, Movie S2). We adopted a closed-loop design for triggering light inactivation of VTA neurons on each trial when the female came in proximity to the pup, thus targeting the peak of VTA activity for suppression. We used real-time position tracking to initiate delivery of continuous light to the VTA while the mouse’s snout was in a circular (6.6 cm diameter) trigger zone around the pup’s starting location (Figure 6d). Prior to closed-loop behavior, all subjects were tested with a standard retrieval assay. There was no significant difference between the performance of stGtACR-expressing and EGFP-expressing mice on this pre-test (Figure 6e). Pre-test performance was also not correlated with their performance during light inactivation trials on P0 (Figure 6f).

stGtACR-expressing mice took significantly longer to retrieve pups to the nest across P0 – 3 (Figure 6g). Although they showed a trend of faster performance over 4 days of testing, they were still slower than the GFP-expressing mice on the last day (Figure 6i, Figure S6). This was because the stGtACR-expressing mice exhibited a longer latency to approach the pup (Figure 6h, 6j, Figure S6). Moreover, the stGtACR-expressing mice exhibited reduced approach velocity compared to the GFP-expressing mice (Figure S6d–e). However, the duration and the velocity in the trigger zone were indistinguishable in the two groups of mice (Figure 6k, Figure S6f), suggesting that the optogenetic inactivation of VTA DA neurons did not acutely affect the motor function of the mice.

Activity of VTA DA neurons is required for reinforcement but not execution of pup retrieval

To further clarify the role of VTA DA neuron activity in reinforcement learning of maternal behavior, we modified our optogenetic inactivation paradigm such that light inactivation of VTA DA neurons was triggered on alternating trials (Figure 7a). As above, stGtACR-expressing mice exhibited indistinguishable pre-test performance index from that of GFP-expressing mice (Figure 7b), and their pre-test performance was uncorrelated with their retrieval performance on P0 (Figure 7c).

Figure 7: — (a) Experimental design similar to Figure 6, but VTA neurons were inhibited on every other trial (Movie S2).

(b) A barplot of results of a pre-test that showed no significant difference in performance before training (n = 7 and 8 for stGtACR-expressing and EGFP-expressing mice, Mann-Whitney test, p > 0.05).

(c) Scatterplot showing no relationship between pre-test performance and the effects of inactivating VTA DA neurons on P0.

(d-f) Plots of retrieval latency (d) (two-way ANOVA with Sidak correction, main factor (day): p < 0.001, main factor (mouse type): p < 0.001, *** p < 0.001) , approach latency (e) (two-way ANOVA with Sidak correction, main factor (day): p < 0.0001, main factor (mouse type): p < 0.001, *** p < 0.001) , and time in trigger zone (f) (two-way ANOVA with Sidak correction, main factor (day): p < 0.0001, main factor (mouse type): p > 0.05) across days comparing stGtACR-expressing (blue) and GFP-expressing (black) mice.

(g-i) Same as (d-f) but with light on and light off trials plotted separately (two-way ANOVA with Tukey correction, main factor (day): p < 0.001 for (g-h) and p < 0.0001for (i), main factor (mouse type): p < 0.001 for (g-h) and p > 0.05 for (i), **p < 0.01, ***p < 0.001, * represents significant difference in opto on vs both control on and off groups, # represents significant difference in opto off vs both control on and off groups).

Data are represented as mean ± SEM.

See also Movie S2 and Figure S7.

stGtACR-expressing mice were initially dramatically slower to return pups to the nest and took more trials to achieve the same performance when compared with EGFP-expressing controls (Figure 7d, Movie S2). This difference was predominantly due to delay in approaching the pup. stGtACR-expressing mice took significantly longer to approach the pup than EGFP-expressing mice (Figure 7e). There was no difference in time spent in the trigger zone retrieving the pup (Figure 7f). Interestingly, the retrieval velocity in the stGtACR-expressing mice was slower than the EGFP-expressing mice and the velocity of both groups became slower over days, probably due to the increasing weight of the pup (Figure S7d). On the other hand, the approach velocity of both groups of mice became faster over days, but the stGtACR-expressing mice approached the pup more slowly than the EGFP-expressing mice (Figure S7e), suggesting the light inhibition of dopaminergic neurons in the VTA impairs their performance and motivation in pup retrieval. Similar to the mice that received light stimulation in every trial, there were no differences in their duration and velocity in the trigger zone (Figure 7f, Figure S7f).

Comparing trials with and without light inhibition, stGtACR-expressing mice exhibited indistinguishable overall retrieval latency, latency to approach, and time in the trigger zone (Figure 7g–i). Moreover, the retrieval velocity, approach velocity and velocity in the trigger zone were comparable in stGtACR-expressing mice, regardless of the light inactivation (Figure S7g–i). Thus, light inhibition on a given trial was irrelevant to performance on that trial. We conclude that impaired retrieval performance reflects the cumulative effect of VTA inactivation on reward history, not acute interference with behavior. Our measurements of VTA DA signals are consistent with RPE, which reinforces future behavior, rather than motivation or motor invigoration, which acutely modifies current behavior. Moreover, the pattern of results indicates that pup retrieval learning is largely driven by learning the value of the behavior, rather than by learning aspects of its motor performance.

DISCUSSION

Dopamine is a central participant in several aspects of motivated behavior. It is acutely important for motivating the initiation of goal-directed behavior and for invigorating action, yet it also functions as a reward signal that cumulatively reinforces future behaviors. This aspect of DA is widely believed to rely on reward prediction. A number of recent studies have outlined the circuitry of social reward, and the mesolimbic DA pathway is a crucial hub in this network ^2–7. Solie et al. ⁷ in particular showed that many dopaminergic VTA neurons are active during social encounters between adults, and that this activity bears features of reward prediction. Here we were able to perform a trial-by-trial analysis of pup retrieval, a repetitive and stereotyped behavior that is fundamental to maternal care. These features of the behavior allowed us to observe the dynamics of the underlying neural activity over many trials with a common event structure.

The DA neuron responses we observed during retrieval were inconsistent with acute motivation or motor invigoration for several reasons: First, the signals didn’t begin until the female made contact with the pup. Second, the magnitude of signals on a given trial were inversely correlated with measures of performance and vigor for that trial. Third, while inactivation of VTA DA neurons at pup contact did dramatically slow acquisition and improvement of retrieval, there was little to no acute effect on ongoing behavior. Two previous studies argued, based on optogenetic manipulations of hypothalamic inputs to VTA, that VTA is important for motivating parental behavior ^28,44. Nevertheless, in our direct measurement of VTA DA neuron signals, we found no evidence of activity that was consistent with acutely enhancing motivation. Esr1-expressing hypothalamic neurons that project to VTA reportedly preferentially target non-DA VTA neurons, which could produce unexpected indirect effects ²⁸. Our results also don’t exclude the likelihood that VTA DA neurons may contribute to maternal motivation on longer timescales ^13,28,44.

We also made several observations that argue that VTA DA neurons signal RPE and reinforce future pup retrievals. First, the signals were positively correlated with performance on the next trial, suggesting they were reinforcing for retrieval. Second, as pup contact became more predictable, pup contact responses diminished and responses to pup-predicting stimuli (e.g. the door opening) grew stronger. Third, the signals were prominently modulated by reward expectation and appeared to reflect the difference from expected reward, not reward itself. An important caveat to this conclusion is that it is based on population level fiber photometry measurements, which may obscure some of the well-described functional heterogeneity among VTA outputs e.g. ^{13,17,45,46,47}. Nevertheless, our results suggest these are robust properties of this pathway with respect to pup retrieval and social reward, and they are consistent with prior studies of instructed behavior e.g. ^48,49. The reward associated with pup retrieval may be partly motivated by avoiding potentially aversive pup cries, however access to pups is very clearly a positive reinforcer ^20–24 and mimics responses to treats (Figure S2).

In our cued retrieval task, we were surprised to find VTA responses to the CS+, which is a reward-predicting cue, were generally weak and inconsistent across mice. Instead, we found that opening of the door opening acquired stronger responses over repeated trials (Figure 4). This could be because the door is a more proximal predictor of the unconditioned stimulus. Interestingly, although the sound was identical for all trials, the responses of VTA DA neurons to the door increased only for the CS+ trials and not for the CS− trials (Figure 4e). This difference argues that despite the lack of VTA responses to the cue, mice behaved as if they understood its significance. Indeed, this is also evident in the higher VTA activity just prior to the door opening in CS+ catch trials as compared to CS− trials. Perhaps with sufficient trials, the CS+ tone would acquire a response in VTA DA neurons.

The mesolimbic DA pathway accesses several downstream targets that could alter maternal behavior. One likely target is nucleus accumbens (NA), which is crucial for reward processing broadly. This is consistent with our results suggesting that VTA modulates parental behavior through a reinforcement learning mechanism, and is corroborated by reports of DA release in NA during pup interactions ^32–39. Numan and colleagues ^50,51 proposed a model in which inputs from VTA and basal amygdala converge on NA and synergize such that responses to pup-related sensory cues in NA are enhanced.

Several neuromodulatory systems have been implicated in facilitating the emergence of maternal retrieval behavior including the oxytocin system ^52,53, noradrenaline system ^54,55, and dopamine system ^25,27–31. Our results are broadly consistent with the known involvement of DA in maternal behavior. However, we significantly expand this understanding by revealing the subsecond temporal structure of VTA activity during maternal care and how it adapts on each trial to dynamically regulate behavior. Importantly, we also propose a mechanism for how DA can cumulatively shape patterns of interactive behavior that is consistent with a known role of DA reinforcement learning. Finally, we established the necessity of that mechanism using contact triggered optogenetic manipulation of VTA DA neuron. VTA DA signals likely influence maternal behavior by increasing the probability and quality of future retrieval, rather than by acutely motivating or invigorating ongoing retrieval.

These findings complement recent work revealing that VTA DA neurons also play a role in other spontaneous natural behaviors including vocal imitation in juvenile songbirds ^56–58 and response selection during social defeat ⁵⁹. More broadly, our results are consistent with a model of intraspecific interaction in which social decisions are cumulatively influenced by the motivation to maximize rewards obtained through social contact. Results in nonhuman primates reveal the quantifiably high value and fungibility of social stimuli in terms of non-social reward currency such as juice ^60,61. Therefore, this model may be a useful framework for understanding social decision making in diverse species, including humans.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Stephen Shea (sshea@cshl.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

All data reported in this study are available from the lead contact upon request.

All original code has been deposited at Zenodo and is publicly available now. The DOI is listed in the key resources table.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken anti-TH	Abcam	Cat #76442; RRID: AB_1524535
Rabbit anti-GFP	Invitrogen	Cat# A6455; RRID: AB_221570
Goat anti-chicken Alexa Fluor 594	Invitrogen	Cat# A11042; RRID: AB_2534099
Goat anti-chicken Alexa Fluor 488	Invitrogen	Cat# A11039; RRID: AB_2534096
Goat anti-rabbit Alexa Fluor 488	Invitrogen	Cat# A11008; RRID: AB_143165
Bacterial and virus strains
AAV-syn-FLEX-jGCaMP7f-WPRE	Addgene	Cat# 104492-AAV9
AAV-hSyn1-SIO-stGtACR2-FusionRed	Addgene	Cat# 105677-AAV1
AAV-synP-DIO-EGFP-WPRE-hGH	Addgene	Cat# 100043-AAV9
Chemicals, peptides, and recombinant proteins
DirectPCR	Viagen	Cat# 402-E
Proteinase K	Invitrogen	Cat# 2036707
Euthasol	Virbac	Cat# 200-071
4% paraformaldehyde	FD Neurotechnologies	Cat# PF101
OCT compound	Sakura	Cat# 4583
Normal goat serum	Vector Laboratories	Cat# S-1000-20
Triton X-100	Sigma	Cat# T8787
Ketamine	Ketaset	Cat# NDC 54771-2013
Xylazine	AnaSed	Cat# NDC 59399-110-20
Meloxicam	Metacam	Cat# NDC 0010-6013-01
Baytril	Bayer	Cat# 08713254-186599
Metabond	C&B	Cat# S380
Critical commercial assay
GoTaq Green Master Mix	Promega	M7123
Experimental models: Organisms/strains
DAT-Cre (Slc6a3tm1.1(cre)Bkmn)	The Jackson Laboratory	Cat# 006302
CBA/CaJ	The Jackson Laboratory	Cat# 000654
Software and algorithms
Zen	Zeiss	https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html
MATLAB R2019b	MatWorks	https://www.mathworks.com/products/matlab.html
Spike2	Cambridge Electronic Design	https://ced.co.uk/products/spkovin
Custom analysis code		http://doi.org/10.5281/zenodo.7377063
Other
Microtome	Leica	SM 2010R
Confocal microscope	Zeiss	LSM 710
Optic fibers for fiber photometry	Thorlabs	Cat# CFMLC12U
Dual fiber cannulae for optogenetics	Doric lenses	Cat# B280-2013_6
Industrial camera BlackFly S	Teledyne FLIR	Cat# FL3-U3-13Y3M-C
Step motor	McMaster-Carr	Cat# 6627T33
Step motor driver	McMaster-Carr	Cat# 6627T41
Arduino Uno R3	Arduino	Cat# A000066
Pulse Pal v2	Sanworks	Cat# 1102
Amplifier	Brownlee Precision	Cat# 410
TDT electrostatic speaker and driver	TDT	Cat# ES1 and ED1
Patch cord for fiber photometry	Doric Lenses	Cat# P99414-01
Mating sleeves	Thorlabs	Cat# ADAL1
470nm and 565nm LED light sources	Thorlabs	Cat# M470F3 and M565F3
LED driver	Thorlabs	Cat# LEDD1B
Photoreceivers	Newport	Cat# 2151
NIDAQ boards	National Instruments	Cat# USB-6211 and USB-6001
Rotary joint for optogenetics	Doric lenses	Cat# FRJ_1×2i_FC-2FC
473nm laser and laser driver	OEM Laser Systems	Cat# PSU III LED
1.0 MΩ tungsten microelectrode	Microprobe	Cat# WE30031.0A3
AC differential amplifier (model 1800)	A–M Systems	Cat# 700000

Open in a new tab

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at Cold Spring Harbor Laboratory and performed in accordance with the National Institutes of Health’s Guide for the care and use of laboratory animals. Mice were maintained in a 12h light/12h dark cycle with ad libitum access to food and water. Test animals were adult DAT-Cre female mice over 8 weeks old (Slc6a3tm1.1(cre)Bkmn; The Jackson Laboratory, 006302). Pups and dams used in the experiments were from CBA/CaJ pairs (The Jackson Laboratory, 000654). For optogenetics experiments, littermates were used for the control and the test groups.

Genotyping

After weaning on P21, the offspring of DAT-Cre animals were genotyped by taking an ear punch. Each sample was first digested with 100 µL DirectPCR (Viagen, 402-E) and 1 µL Proteinase K (Invitrogen, 2036707) at 60 °C for 4 hours, and then the Proteinase K was inactivated at 95 °C for 45 minutes. The PCR solution for each sample contained 12.5 µL GoTaq Green Master Mix (Promega, M7123), 1.5 µL common forward primer, 1.5 µL reverse primer for the mutant or the wild-type allele, 2 µL genomic DNA, and 7.5 µL dH2O. Primer sequences, PCR reactions and electrophoresis were prepared according to a protocol from the Jackson Laboratory.

Histology

Animals were deeply anesthetized with a lethal dose of Euthasol (Virbac, 200-071) via intraperitoneal injection, and were subsequently transcardially perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA; FD Neurotechnologies, PF101) at a flow rate of 5 mL/min. Brains were post-fixed in 4% PFA overnight at 4 °C and then transferred to a 30% sucrose solution in PBS. Brains were embedded with OCT compound (Sakura, 4583) and sectioned frozen at 50 µm on a sliding microtome (Leica, SM 2010R). For immunostaining, free-floating sections were washed with PBS 3 times and blocked in 5% normal goat serum (NGS, Vector Laboratories, S-1000-20) and 0.1% Triton X-100 (Sigma, T8787) at room temperature (RT) for 1 hour. Sections were incubated overnight with primary antibodies in a solution containing 0.5% NGS and 0.1% Triton X-100 at 4 °C. The next day, sections were incubated for 2 h with secondary antibodies in a solution containing 0.5% NGS and 0.1% Triton X-100 at room temperature. After washing 3 times with 1X PBS, sections were mounted on coverslips with Fluoromount-G (SouthernBiotech, 0100-01). Primary antibodies including chicken anti-TH antibody (Abcam, 76442) and rabbit anti-GFP antibody (Invitrogen, A6455) were used. Secondary antibodies including Alexa Fluor 594 goat anti-chicken (Invitrogen, A11042), Alexa Fluor 488 goat anti-chicken (Invitrogen, A11039) and Alexa Fluor 488 goat anti-rabbit (Invitrogen, A11008) were used.

Imaging and image analysis

Brain sections were imaged with a Zeiss LSM 710 confocal microscope. Sections that contained VTA were imaged and analyzed with Zeiss Zen software.

Viruses

We used the following commercially-available viruses: AAV-syn-FLEX-jGCaMP7f-WPRE (2.8 ×10¹³ GC/mL, Addgene, 104492-AAV9), AAV-hSyn1-SIO-stGtACR2-FusionRed (1×10¹³ GC/mL, Addgene,105677-AAV1), AAV-synP-DIO-EGFP-WPRE-hGH (0.9×10¹³ GC/mL, Addgene,100043-AAV9). All viruses were aliquoted immediately on delivery and stored at −80 °C until use.

Stereotaxic surgery

Animals were injected intraperitoneally with a mixture of ketamine (100 mg/kg, Ketaset, NDC 54771-2013) and xylazine (5 mg/kg AnaSed, NDC 59399-110-20), and with meloxicam (1 – 2 mg/kg, Metacam, NDC 0010-6013-01) and Baytril (10 mg/kg, Bayer, 08713254-186599) subcutaneously prior to surgeries. Animals were placed in a stereotaxic device and maintained at a plane of anesthesia with 1 – 2% isoflurane mixed with oxygen at a flow rate of 1 – 2 L/min. For fiber photometry experiments, 200 – 800 nL Cre-dependent jGCaMP7f AAV was unilaterally injected into the VTA at 10 nL/min. Standard coordinates used for injections were as follows: AP −3.00 and ML +0.50 relative to bregma, and DV −4.10 relative to the brain surface) Optical fibers (200 µm dia., NA 0.39) from Thorlabs (CFMLC12U) were implanted above VTA (AP: −3.00, ML: +0.50, DV: −4.00 from brain surface) by slowly advancing at a speed of 1 mm/min. For optogenetics experiments, 120 – 200 nL Cre-dependent stGtACR2 and EGFP AAV was bilaterally injected into the VTA (AP: −3.00, ML: −0.50 and +0.50, DV: −4.10 from brain surface). Dual fiber cannulae (200 µm dia., NA 0.37 from Doric Lenses, B280-2013_6) were implanted above VTA (AP: −3.00, ML: −0.50/+0.50, DV: −4.00 from brain surface). Real time tracking of the animal’s position during optogenetics experiments was achieved by minature infrared LEDs secured to the skull with dental cement (Metabond, C&B, S380). Animals were allowed to recover from surgery for 3 weeks before commencing experiments.

Behavioral annotation and tracking

All behavior sessions were video recorded from above at 30 frames/s in a dark, sound-attenuated chamber under infrared illumination. Video data were acquired to a computer by either a Logitech webcam in fiber photometry experiments or an industrial camera (Teledyne FLIR, BlackFly S) in optogenetic experiments. Behavior sessions were conducted during the animals’ light phase (9:00h – 19:00h).

In pup retrieval behavior assays for fiber photometry experiments (Figure 1 – 3), the test animals were habituated for 10 minutes in their home cages with a light and flexible optical fiber cable connected to their implant. We ran three cohorts of mice: (1) Co-housed (CH) females were virgin DAT-Cre female mice that were introduced to the cage of pregnant, primiparous CBA/CaJ female 1 – 5 days before the dam gave birth. At that time, the sire was removed and the virgin and dam were co-housed through P5. Each postnatal testing day (P0 – P5), the CH female was tested alone with 5 scattered pups for 5 minutes in each of 3 sessions; (2) Non co-housed (NCH) females were virgin DAT-Cre female mice that lived in their home cages with other virgin females only throughout the experiment. Each day from P0 through P5, the NCH virgin was tested alone with 5 scattered pups for 5 minutes in 1 session; (3) High experience, non co-housed (HE/NCH) females were virgin DAT-Cre female mice that each day from P0 through P5, the HE/NCH virgins were tested alone with 8 scattered pups for 5 minutes for each of 3 sessions. Retrieval latency index was calculated as described ⁶².

r e t r i e v a l l a t e n c y i n d e x = \frac{[(t_{1} - t_{0}) + (t_{2} - t_{0}) + \dots + (t_{n} - t_{0})]}{n \times L}

where n = number of pups scattered in the cage initially, t₀ = time of the start of a trial, t_n = time of retrieving the nth pup, L = 300 s. We used performance index, defined as (1 – retrieval latency index) ⁶² as a metric to evaluate the retrieval performance in mice (Figure 2, 6–7).

For the cued retrieval task with fiber photometry (Figure 4 – 5), virgin DAT-Cre mice were co-housed with the CBA/CaJ dams through P8. Each day the virgin females were tested alone in their home cages. They were habituated for 5 – 10 minutes and then delivered one treat per daily session at a random time to measure neural responses to an unexpected treat. Subsequently, a chamber with a sliding door controlled by a stepper motor (McMaster-Carr, 6627T33), a stepper motor driver (McMaster-Carr, 6627T41), Arduino Uno R3 (Arduino, A000066) and Pulse Pal vs (Sanworks, 1102) was placed in the home cage. Depending on the trial type, a pup was manually placed in the accessible or the inaccessible area inside the chamber. 40 trials were run each day except for P8 on which either 40 or 80 trials were run. Five seconds after a trial started, a CS+ (8 kHz tone) cue or a CS− cue (16 kHz tone) with a duration of 0.5 s or 2 s and a volume of 70 dB at the animal’s head was processed by an amplifier (Brownlee Precision, 410) and presented via an electrostatic speaker and its driver (TDT, ES1 and ED1). The door began to open 1 s after the tone stopped playing, and it took 2 s for the door to fully open. This allowed the animal to enter the chamber to search for a pup. The daily schedule of trials can be found in Figure S4a–b.

For optogenetics experiments (Figure 6 – 7), virgin DAT-Cre mice were co-housed with pregnant CBA/CaJ dams until P0, and afterwards they were co-housed with their littermates. At the beginning of P0, the DAT-Cre virgins were subjected to a pup retrieval assay (pre-test) for 5 minutes in which 5 pups were scattered in their home cages. Each day from P0 through P3, they were habituated with a dual optical fiber attached to their implant in their home cage for 10 minutes, followed by 20 trials with alternating light off and on. In each trial, a pup was placed in the Region of Interest (ROI) for the test animal to retrieve.

Behaviors were annotated offline using the open source software package BORIS. Pup contact was defined as the first frame on which the animal put its mouth onto the pup at the end of approach in a pup retrieval trial. Retrieval was defined as the first frame on which the animals lifted the pup from the ground. The start of pup approach was defined as the moment animals walked out of the nest for pup retrieval. The end of pup approach was defined as the moment animals made contact with a pup. Snack eating was defined as the first frame that the animals bit on the snack in its hands.

To track animal position in the fiber photometry experiments, a deep learning-based method, DeepLabCut ⁴¹ was used. Distance traveled during approach in a pup retrieval trial was calculated based on the coordinates of the midpoint between the ears. Velocity was computed as the distance traveled between two consecutive frames multiplied by the frame rate, followed by convolution with a boxcar kernel of 7 points. Mean velocity of approach was calculated as the average of the convoluted velocity during approach. Head orientation was characterized as the cosine value of the angle between two vectors: position of the pup to the left and right ears respectively. Probability of direct approach was measured by the percentage of frames with cosine values greater than or equal to 0.8 during pup approach. For videos of fiber photometry experiments, 1 pixel is equal to 0.07 cm. In optogenetics experiments, animal position was tracked in real time by a computer with a camera and running Bonsai ⁶³ in which an image processing pipeline was used to extract the position of the LEDs on animals’ heads. For videos of optogenetics experiments, 1 pixel is equal to 0.06 cm.

Fiber photometry

Fiber photometry was conducted as described ⁵⁴. Briefly, the implant was coupled to a patch cord (Doric Lenses, P99414-01) through a mating sleeve (Thorlabs, ADAL1). Two LED light sources (470 nm and 565 nm; Thorlabs, M470F3 and M565F3;) were sinusoidally modulated by LED drivers (Thorlabs, LEDD1B) 180 degrees out of phase at a frequency of 211 Hz. Light power was adjusted to 30 µW at the start of each recording session. The emitted light passed through the patch cord, a focusing lens and dual edge dichroic mirrors to split it to separate photoreceivers (Newport, 2151). Signals were sampled at 6.1 kHz and acquired by NIDAQ boards (National Instruments, USB-6211 and USB-6001).

Data were analyzed with custom MATLAB (MathWorks) software. Briefly, the peaks of the two signals were extracted to achieve an effective sampling rate of 211 Hz and the data were filtered with a low-pass Butterworth filter with a corner frequency of 15 Hz. Signals were fitted with a double exponential function that was subtracted to correct for photobleaching. Next, we used a robust regression algorithm to fit a linear function for predicting the activity-independent component of the green fluorescence from the red fluorescence. The predicted green signal was subtracted and its mean was taken as baseline. ΔF/F was then calculated by dividing the residual green signal by the baseline.

The ΔF/F traces were converted to a Z-score using the mean and standard deviation of all signals recorded from a given subject. The Z-scored ΔF/F traces from individual trials were aligned to specific behaviors to generate heatmaps and compute mean fluorescence. When aligning the ΔF/F signals to specific behaviors, baseline fluctuations were removed by subtracting the mean signal in a time window extending 2 s prior to the event window. Responses to a given behavior during free retrieval (Figure 1 – 3) were quantified as area under curve (AUC) for 4 s after the event. Responses to retrieval in Figure 4 – 5 were quantified as AUC for −2 – 2s around the events. Responses to snack eating (Figure S2) were measured as AUC for −4 – 4s around the events. To compare the changes of retrieval responses over days in different groups of animals (Figure 2 – 3), the AUC data were normalized to those on P0 within each animal. In the cued retrieval task, mean fluorescence was calculated for the designated window relative to the mean activity at the start of the trial before tone presentation. Linear time warping was used to align data to both the door opening and retrieval by aligning all traces to retrieval and then resampling to normalize each entire trial to the median time between the door opening and retrieval.

Closed-loop optogenetics

The bilateral implant was coupled to an optical fiber that connected with a rotary joint (Doric Lenses, FRJ_1×2i_FC-2FC) and a 473-nm laser (OEM Laser Systems, PSU III LED). The laser driver was controlled by a Pulse Pal (Sanworks, v2) that received commands from Bonsai ⁶³. The software also tracked the head position of the animal in real time with the aid of a miniature IR LED affixed to its head. Constant laser stimulation was delivered when the animal’s snout was detected inside the ROI before the pup was returned to the nest. The light power was adjusted to ~10 mW per hemisphere immediately before and after each experiment. For analyzing the behaviors, retrieval and approach latencies were capped at 300 s. Coordinates of the mouse from live tracking in Bonsai were convoluted with a boxcar kernel of 7 points, followed by a third-order one-dimensional median filter to smoothen the data. Approach velocity was calculated as distance from the approach starting position to the pup at the fixed position divided by the approach duration. Retrieval velocity was calculated as distance from the pup at the fixed position to the nest re-entry point divided by the retrieval duration.

In vivo extracellular electrophysiology

Virgin female DAT-Cre mice were injected unilaterally in VTA with AAV driving expression of Cre-dependent stGtACR as described above. After at least three weeks of recovery time, mice were acutely anesthetized with 1 – 2% isoflurane. Custom optrodes were constructed by gluing a 200 µm optical fiber to the shaft of a tungsten microelectrode (1.0 MΩ; Microprobe, WE30031.0A3). Extracellular neural data were recorded with an AC differential amplifier (Model 1800; A–M Systems, 700000). Signals were bandpass filtered between 300 and 3000 Hz and digitally acquired to disk at 10 kHz with software and hardware from Cambridge Electronic Design (Power 1401, Spike2) for later spike sorting and analysis. Neurons encountered in the vicinity of VTA were tested with 5 – 15 10 mW pulses of constant 473 nm light (OEM Laser Systems, PSU III LED) at 30 s intervals.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics

Figure 2d: two-way ANOVA with Tukey’s multiple comparison test. H₀: the normalized AUC of retrieval responses do not differ across days of retrieval experience (factor A). F (5, 69) = 5.472, p = 0.0003. H₀: the normalized AUC of retrieval responses do not differ between three groups of mice (factor B). F (2, 69) = 11.32, p < 0.0001. H₀: there is no interaction between factor A and B. F (10, 69) = 5.472, p = 0.4235. For multiple comparison test, p = 0.0105 (CH vs NCH on P2), p = 0.0066 (CH vs NCH on P3).

Figure 2e: two-way ANOVA with Tukey’s multiple comparison test. H₀: the performance index does not differ across days of retrieval experience (factor A). F (5, 69) = 1.739, p = 0.1373. H₀: the performance index does not differ between three groups of mice (factor B). F (2, 69) = 19.05, p < 0.0001. H₀: there is no interaction between factor A and B. F (10, 69) = 0.4294, p = 0.9273. For multiple comparison test, p = 0.0008 (CH vs NCH on P0), p = 0.0189 (CH vs NCH on P1), p = 0.0214 (CH vs NCH on P3), p = 0.0289 (NCH vs HE/NCH on P0).

Figure 2f–h: Spearman’s correlation. (f) r = −0.54, p = 1.7 × 10⁻³, n = 38; (g) r = −0.43, p = 6.8 × 10⁻³, n = 31; (h) r = −0.56, p = 0.018, n = 18.

Figure 3c: two-way ANOVA with Tukey’s multiple comparison test. H₀: the approach velocity does not differ across days of retrieval experience (factor A). F (5, 69) = 10.14, p < 0.0001. H₀: the approach velocity does not differ between three groups of mice (factor B). F (2, 69) = 27.39, p < 0.0001. H₀: there is no interaction between factor A and B. F (10, 69) = 0.4984, p = 0.8854. For multiple comparison test, p = 0.0044 (CH vs HE/NCH on P0), p = 0.0119 (CH vs HE/NCH on P1), p = 0.0234 (CH vs HE/NCH on P3), p = 0.0027 (NCH vs HE/NCH on P0), p = 0.0077 (NCH vs HE/NCH on P1), p = 0.0011 (NCH vs HE/NCH on P2), p = 0.0228 (NCH vs HE/NCH on P5).

Figure 3e: two-tailed Wilcoxon matched pairs signed rank test, return velocity: p = 0.23, n = 16 mice; approach velocity: p = 4.4 × 10⁻⁴, n = 16 mice; change in approach velocity: p = 6.1 × 10⁻⁴, n = 10 mice.

Figure 3g–i: Pearson’s correlation. (g) r = 0.13, p = 6.7 × 10⁻³, n = 433; (h) r = 0.29, p = 1.6 × 10⁻⁴, n = 155; (i) r = 0.31, p = 1.1 × 10⁻¹⁰, n = 411.

Figure 4g: one-way repeated measures ANOVA with Tukey’s multiple comparisons test. ANOVA results: F (2, 36) = 99.1, p < 0.0001; multiple comparisons test results: p < 0.0001 (baseline vs. door), p < 0.0001 (baseline vs. retrieval), p < 0.0001 (door vs. retrieval)

Figure 4i: paired t test comparisons with Bonferroni correction; pre door early vs. late: p = 0.051, door early vs. late: p = 1.2 × 10⁻⁵, retrieval early vs. late: p = 0.018)

Figure 5c: paired t test comparisons with Holm-Bonferroni correction; entry CS+ vs. CS− catch: p = 0.025, pre-entry CS− vs. CS+ catch: p = 0.0087, post-entry CS+ vs. CS− catch: p = 0.016)

Figure 6e: two-tailed Mann-Whitney test, p = 0.4318, n = 7 and 5 for opto and control groups respectively.

Figure 6f: Pearson’s correlation. r = −0.33, p = 0.46, n = 7 for the opto group (blue); r = 0.13, p = 0.82, n = 5 for the control group (black).

Figure 6g: two-tailed Mann-Whitney test, p = 0.0025, n = 7 and 5 for opto and control groups respectively.

Figure 6h: two-tailed Mann-Whitney test, p = 0.0025, n = 7 and 5 for opto and control groups respectively.

Figure 6i: two-way ANOVA with Sidak’s multiple comparison test. H₀: the retrieval latency does not differ across days of retrieval experience (factor A). F (3, 30) = 2.429, p = 0.0847. H₀: the retrieval latency does not differ between the two groups of mice (factor B). F (1, 10) = 12.00, p = 0.0061. H₀: there is no interaction between factor A and B. F (3, 30) = 0.9697, p = 0.4199. For multiple comparison test, p = 0.0144 on P0, p = 0.0057 on P1.

Figure 6j: two-way ANOVA with Sidak’s multiple comparison test. H₀: the approach latency does not differ across days of retrieval experience (factor A). F (3, 30) = 1.901, p = 0.1508. H₀: the approach latency does not differ between the two groups of mice (factor B). F (1, 10) = 11.66, p = 0.0066. H₀: there is no interaction between factor A and B. F (3, 30) = 0.9534, p = 0.4274. For multiple comparison test, p = 0.0152 on P0, p = 0.0068 on P1.

Figure 6k: two-way ANOVA. H₀: the duration in ROI does not differ across days of retrieval experience (factor A). F (3, 30) = 0.5920, p = 0.6251. H₀: the duration in ROI does not differ between the two groups of mice (factor B). F (1, 10) = 0.5974, p = 0.4575. H₀: there is no interaction between factor A and B. F (3, 30) = 2.836, p = 0.0528.

Figure 7b: two-tailed Mann-Whitney test, p = 0.0541, n = 7 and 8 for opto and control groups respectively.

Figure 7c: Pearson’s correlation. R = −0.35, p = 0.45, n = 7 for the opto group (blue); r = −0.08, p = 0.86, n = 8 for the control group (black).

Figure 7d: two-way ANOVA with Sidak’s multiple comparison test. H₀: the retrieval latency does not differ across days of retrieval experience (factor A). F (3, 39) = 11.64, p < 0.0001. H₀: the retrieval latency does not differ between the two groups of mice (factor B). F (1, 13) = 26.75, p = 0.0002. H₀: there is no interaction between factor A and B. F (3, 39) = 7.417, p = 0.0005. For multiple comparison test, p < 0.0001 on P0, p = 0.0004 on P1.

Figure 7e: two-way ANOVA with Sidak’s multiple comparison test. H₀: the approach latency does not differ across days of retrieval experience (factor A). F (3, 39) = 9.590, p < 0.0001. H₀: the approach latency does not differ between the two groups of mice (factor B). F (1, 13) = 25.66, p = 0.0002. H₀: there is no interaction between factor A and B. F (3, 39) = 7.365, p = 0.0005. For multiple comparison test, p < 0.0001 on P0, p = 0.0005 on P1.

Figure 7f: two-way ANOVA. H₀: duration in ROI does not differ across days of retrieval experience (factor A). F (3, 39) = 17.63, p < 0.0001. H₀: the duration in ROI does not differ between the two groups of mice (factor B). F (1, 13) = 4.582, p = 0.0518. H₀: there is no interaction between factor A and B. F (3, 39) = 1.448, p = 0.2438.

Figure 7g: two-way ANOVA with Tukey’s multiple comparison test. H₀: the retrieval latency does not differ across days of retrieval experience (factor A). F (3, 78) = 19.32, p < 0.0001. H₀: the retrieval latency does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 14.57, p < 0.0001. H₀: there is no interaction between factor A and B. F (9, 78) = 4.281, p = 0.0002. For multiple comparison test, p < 0.0001 (opto on vs control on, opto on vs control off, opto off vs control on and opto off vs control off on P0), p = 0.0083 (opto on vs control on on P1), p = 0.0087 (opto on vs control off on P1), p = 0.0001 (opto off vs control on and opto off vs control off on P1).

Figure 7h: two-way ANOVA with Tukey’s multiple comparison test. H₀: the approach latency does not differ across days of retrieval experience (factor A). F (3, 78) = 15.75, p < 0.0001. H₀: the approach latency does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 13.76, p < 0.0001. H₀: there is no interaction between factor A and B. F (9, 78) = 4.220, p = 0.0002. For multiple comparison test, p < 0.0001 (opto on vs control on, opto on vs control off, opto off vs control on and opto off vs control off on P0), p = 0.0119 (opto on vs control on on P1), p = 0.0112 (opto on vs control off on P1), p = 0.0002 (opto off vs control on and opto off vs control off on P1).

Figure 7i: two-way ANOVA. H₀: the duration in ROI does not differ across days of retrieval experience (factor A). F (3, 78) = 25.78, p < 0.0001. H₀: the duration in ROI does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 2.485, p = 0.0830. H₀: there is no interaction between factor A and B. F (9, 78) = 1.035, p = 0.4206.

Figure S1b: two-way ANOVA with Tukey’s multiple comparison test. H₀: the AUC of pup contact response does not differ across days of retrieval experience (factor A). F (5, 62) = 4.183, p = 0.0024. H₀: the AUC of pup contact response does not differ between three groups of mice (factor B). F (2, 64) = 5.177, p = 0.0082. H₀: there is no interaction between factor A and B. F (10, 64) = 0.4891, p = 0.8910.

Figure S1e: Pearson’s correlation. r = 0.97, p = 5.8 × 10⁻⁵², n = 31, 38 and 18 for CH, NCH and HE/NCH respectively.

Figure S1f: Kruskal-Wallis test. p = 0.8737, n = 6, 7 and 3 for CH, NCH and HE/NCH respectively.

Figure S2b: one-way ANOVA. F (8, 114) = 1.491, p = 0.1682, n = 15 mice.

Figure S3a: two-way ANOVA. H₀: the time of lifting differs across days of retrieval experience (factor A). F (5, 69) = 1.393, p = 0.2379. H₀: the retrieval onset differs between the three groups of mice (factor B). F (2, 69) = 2.187, p = 0.1200. H₀: there is no interaction between factor A and B. F (10, 69) = 0.368, p = 0.9562.

Figure S3b: two-way ANOVA with Tukey’s multiple comparison test. H₀: the retrieval onset differs across days of retrieval experience (factor A). F (5, 71) = 0.4589, p = 0.8055. H₀: the retrieval onset differs between the three groups of mice (factor B). F (2, 71) = 15.21, p < 0.0001. H₀: there is no interaction between factor A and B. F (10, 71) = 0.3648, p = 0.9578. For multiple comparison test, p < 0.05 (CH vs NCH on P3 and P5).

Figure S3c: two-way ANOVA. H₀: the probability of direct approach differs across days of retrieval experience (factor A). F (5, 69) = 0.8003, p = 0.5533. H₀: the retrieval onset differs between the three groups of mice (factor B). F (2, 69) = 1.112, p = 3348. H₀: there is no interaction between factor A and B. F (10, 69) = 0.4141, p = 0.9354.

Figure S5c: one-way ANOVA. F (2, 15) = 1.491, p = 0.0035, n = 6 cells.

Figure S6d: two-way ANOVA. H₀: the retrieval velocity does not differ across days of retrieval experience (factor A). F (3, 30) = 1.890, p = 0.1526. H₀: the retrieval velocity does not differ between the two groups of mice (factor B). F (1, 10) = 4.765, p = 0.0540. H₀: there is no interaction between factor A and B. F (3, 30) = 0.5667, p = 0.6412.

Figure S6e: two-way ANOVA. H₀: the approach velocity does not differ across days of retrieval experience (factor A). F (3, 30) = 36.97, p < 0.0001. H₀: the approach velocity does not differ between the two groups of mice (factor B). F (1. 10) = 5.336, p = 0.0435. H₀: there is no interaction between factor A and B. F (3, 30) = 0.08940, p = 0.9653.

Figure S6f: two-way ANOVA. H₀: the velocity in ROI does not differ across days of retrieval experience (factor A). F (3, 30) = 1.624, p = 0.2046. H₀: the velocity in ROI does not differ between the two groups of mice (factor B). F (1. 10) = 0.7564, p = 0.4049. H₀: there is no interaction between factor A and B. F (3, 30) = 2.673, p = 0.0652.

Figure S7d: two-way ANOVA with Sidak’s multiple comparison test. H₀: the retrieval velocity does not differ across days of retrieval experience (factor A). F (3, 39) = 1.049, p = 0.3817. H₀: the retrieval velocity does not differ between the two groups of mice (factor B). F (1, 13) = 9.016, p = 0.0102. H₀: there is no interaction between factor A and B. F (3, 39) = 1.029, p = 0.3902. For multiple comparison test, p = 0.0480 on P0, p = 0.0127 on P1.

Figure S7e: two-way ANOVA with Sidak’s multiple comparison test. H₀: the approach velocity does not differ across days of retrieval experience (factor A). F (3, 39) = 51.34, p < 0.0001. H₀: the approach velocity does not differ between the two groups of mice (factor B). F (1, 13) = 25.25, p = 0.0002. H₀: there is no interaction between factor A and B. F (3, 39) = 2.231 p = 0.0999. For multiple comparison test, p = 0.0193 on P0, p = 0.0124 on P1, p < 0.0001 on P2 and P3.

Figure S7f: two-way ANOVA. H₀: the velocity in ROI does not differ across days of retrieval experience (factor A). F (3, 39) = 7.080, p = 0.0007. H₀: the velocity in ROI does not differ between the two groups of mice (factor B). F (1, 13) = 1.867, p = 0.1950. H₀: there is no interaction between factor A and B. F (3, 39) = 0.9780, p = 0.4130.

Figure S7g: two-way ANOVA with Tukey’s multiple comparison test. H₀: the retrieval velocity does not differ across days of retrieval experience (factor A). F (3, 78) = 1.701, p = 0.1736. H₀: the retrieval velocity does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 6.290, p = 0.0024. H₀: there is no interaction between factor A and B. F (9, 78) = 0.7090, p = 0.6989. For multiple comparison test, p = 0.0054 (opto on vs control off on P0), p = 0.0440 (opto off vs control off on P0), p = 0.0066 (opto on vs control off on P1), p = 0.0092 (opto off vs control off on P1), p = 0.0403 (opto on vs control off on P2).

Figure S7h: two-way ANOVA with Tukey’s multiple comparison test. H₀: the approach velocity does not differ across days of retrieval experience (factor A). F (3, 78) = 73.97, p < 0.0001. H₀: the approach velocity does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 15.78, p < 0.0001. H₀: there is no interaction between factor A and B. F (9, 78) = 1.421, p = 0.1938. For multiple comparison test, p = 0.0070 (opto on vs control off on P0), p = 0.0070 (opto off vs control off on P0), p = 0.0119 (opto on vs control on on P1), p = 0.0180 (opto on vs control off on P1), p = 0.0292 (opto off vs control on on P1), p < 0.0001 (opto on vs control on on P2), p < 0.0001 (opto on vs control off on P2), p = 0.0016 (opto off vs control on on P2), p = 0.0002 (opto off vs control off on P2), p = 0.0002 (opto on vs control on on P3), p < 0.0001 (opto on vs control off on P3), p = 0.0022 (opto off vs control on on P3), p = 0.0002 (opto off vs control off on P3).

Figure S7i: two-way ANOVA. H₀: the velocity in ROI does not differ across days of retrieval experience (factor A). F (3, 78) = 11.51, p < 0.0001. H₀: the velocity in ROI does not differ between the two groups of mice with light on/off (factor B). F (3, 26) = 1.289, p = 0.2990. H₀: there is no interaction between factor A and B. F (9, 78) = 0.7678, p = 0.6462.

Supplementary Material

NIHMS1856976-supplement-2.pdf^{(8.4MB, pdf)}

Download video file^{(45.2MB, mov)}

Download video file^{(15.3MB, mp4)}

Highlights.

VTA DA neuron activity during maternal care inversely correlates with performance
Performance of pup retrieval is influenced by the history of DA neuron activity
VTA DA neurons signal reward prediction error for pup contact
Silencing VTA DA neurons affects future performance but not the current trial

ACKNOWLEDGMENTS

The authors would like to thank R. Mooney, S.R. Datta, A. Zador, B. Li, M. Long, and members of the Shea Lab for helpful comments and discussion. This work was supported by grants to SDS from the National Institute of Mental Health (R01MH119250), the C.M. Robertson Foundation, and the Feil Foundation.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

DECLARATION OF INTERESTS

The authors declare no competing interests.

INCLUSION AND DIVERSITY

We support inclusive, diverse, and equitable conduct of research.

SUPPLEMENTAL INFORMATION

Supplemental information includes eight figures and two movies which can be found in the online version of this article.

DATA AND SOFTWARE AVAILABILITY

Raw data and MATLAB code for analysis are available upon request.

REFERENCES

1.Krach S, Paulus FM, Bodden M, and Kircher T (2010). The rewarding nature of social interactions. Front Behav Neurosci 4, 22. 10.3389/fnbeh.2010.00022. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Dolen G, Darvishzadeh A, Huang KW, and Malenka RC (2013). Social reward requires coordinated activity of nucleus accumbens oxytocin and serotonin. Nature 501, 179–184. 10.1038/nature12518. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Hung LW, Neuner S, Polepalli JS, Beier KT, Wright M, Walsh JJ, Lewis EM, Luo L, Deisseroth K, Dolen G, and Malenka RC (2017). Gating of social reward by oxytocin in the ventral tegmental area. Science 357, 1406–1411. 10.1126/science.aan4994. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.McHenry JA, Otis JM, Rossi MA, Robinson JE, Kosyk O, Miller NW, McElligott ZA, Budygin EA, Rubinow DR, and Stuber GD (2017). Hormonal gain control of a medial preoptic area social reward circuit. Nat Neurosci 20, 449–458. 10.1038/nn.4487. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Nardou R, Lewis EM, Rothhaas R, Xu R, Yang A, Boyden E, and Dolen G (2019). Oxytocin-dependent reopening of a social reward learning critical period with MDMA. Nature 569, 116–120. 10.1038/s41586-019-1075-9. [DOI] [PubMed] [Google Scholar]
6.Resendez SL, Namboodiri VMK, Otis JM, Eckman LEH, Rodriguez-Romaguera J, Ung RL, Basiri ML, Kosyk O, Rossi MA, Dichter GS, and Stuber GD (2020). Social Stimuli Induce Activation of Oxytocin Neurons Within the Paraventricular Nucleus of the Hypothalamus to Promote Social Behavior in Male Mice. J Neurosci 40, 2282–2295. 10.1523/JNEUROSCI.1515-18.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Solie C, Girard B, Righetti B, Tapparel M, and Bellone C (2022). VTA dopamine neuron activity encodes social interaction and promotes reinforcement learning through social prediction error. Nat Neurosci 25, 86–97. 10.1038/s41593-021-00972-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Montague PR, Dayan P, and Sejnowski TJ (1996). A framework for mesencephalic dopamine systems based on predictive Hebbian learning. J Neurosci 16, 1936–1947. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Sutton RS, and Barto AG (1998). Reinforcement Learning: An Introduction (MIT Press). [Google Scholar]
10.Schultz W, Dayan P, and Montague PR (1997). A neural substrate of prediction and reward. Science 275, 1593–1599. 10.1126/science.275.5306.1593. [DOI] [PubMed] [Google Scholar]
11.Watabe-Uchida M, Eshel N, and Uchida N (2017). Neural Circuitry of Reward Prediction Error. Annu Rev Neurosci 40, 373–394. 10.1146/annurev-neuro-072116-031109. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Berke JD (2018). What does dopamine mean? Nat Neurosci 21, 787–793. 10.1038/s41593-018-0152-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Hamid AA, Pettibone JR, Mabrouk OS, Hetrick VL, Schmidt R, Vander Weele CM, Kennedy RT, Aragona BJ, and Berke JD (2016). Mesolimbic dopamine signals the value of work. Nat Neurosci 19, 117–126. 10.1038/nn.4173. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Barter JW, Li S, Lu D, Bartholomew RA, Rossi MA, Shoemaker CT, Salas-Meza D, Gaidis E, and Yin HH (2015). Beyond reward prediction errors: the role of dopamine in movement kinematics. Front Integr Neurosci 9, 39. 10.3389/fnint.2015.00039. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Bova A, Gaidica M, Hurst A, Iwai Y, Hunter J, and Leventhal DK (2020). Precisely timed dopamine signals establish distinct kinematic representations of skilled movements. Elife 9. 10.7554/eLife.61591. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.da Silva JA, Tecuapetla F, Paixao V, and Costa RM (2018). Dopamine neuron activity before action initiation gates and invigorates future movements. Nature 554, 244–248. 10.1038/nature25457. [DOI] [PubMed] [Google Scholar]
17.Hughes RN, Bakhurin KI, Petter EA, Watson GDR, Kim N, Friedman AD, and Yin HH (2020). Ventral Tegmental Dopamine Neurons Control the Impulse Vector during Motivated Behavior. Curr Biol 30, 2681–2694 e2685. 10.1016/j.cub.2020.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Ehret G, Koch M, Haack B, and Markl H (1987). Sex and parental experience determine the onset of an instinctive behavior in mice. Naturwissenschaften 74, 47. [DOI] [PubMed] [Google Scholar]
19.Stolzenberg DS, and Champagne FA (2016). Hormonal and non-hormonal bases of maternal behavior: The role of experience and epigenetic mechanisms. Horm Behav 77, 204–210. 10.1016/j.yhbeh.2015.07.005. [DOI] [PubMed] [Google Scholar]
20.Wilsoncroft WE ( 1969). Babies by bar press: maternal behavior in the rat. Behavioral research methods 1, 229–230. [Google Scholar]
21.Fleming AS, Korsmit M, and Deller M (1994). Rat pups are potent reinforcers to the maternal animal: effects of experience, parity, hormones, and dopamine function. Psychobiology 22, 44–53. [Google Scholar]
22.Hauser H, and Gandelman R (1985). Lever pressing for pups: evidence for hormonal influence upon maternal behavior of mice. Horm Behav 19, 454–468. [DOI] [PubMed] [Google Scholar]
23.Dunlap AG, Besosa C, Pascual LM, Chong KK, Walum H, Kacsoh DB, Tankeu BB, Lu K, and Liu RC (2020). Becoming a better parent: Mice learn sounds that improve a stereotyped maternal behavior. Horm Behav 124, 104779. 10.1016/j.yhbeh.2020.104779. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Mattson BJ, Williams S, Rosenblatt JS, and Morrell JI (2001). Comparison of two positive reinforcing stimuli: pups and cocaine throughout the postpartum period. Behav Neurosci 115, 683–694. [DOI] [PubMed] [Google Scholar]
25.Curry T, Egeto P, Wang H, Podnos A, Wasserman D, and Yeomans J (2013). Dopamine receptor D2 deficiency reduces mouse pup ultrasonic vocalizations and maternal responsiveness. Genes Brain Behav 12, 397–404. 10.1111/gbb.12037. [DOI] [PubMed] [Google Scholar]
26.Seip KM, and Morrell JI (2009). Transient inactivation of the ventral tegmental area selectively disrupts the expression of conditioned place preference for pup-but not cocaine-paired contexts. Behav Neurosci 123, 1325–1338. 10.1037/a0017666. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Stolzenberg DS, Zhang KY, Luskin K, Ranker L, Bress J, and Numan M (2010). Dopamine D(1) receptor activation of adenylyl cyclase, not phospholipase C, in the nucleus accumbens promotes maternal behavior onset in rats. Horm Behav 57, 96–104. 10.1016/j.yhbeh.2009.09.014. [DOI] [PubMed] [Google Scholar]
28.Fang YY, Yamaguchi T, Song SC, Tritsch NX, and Lin D (2018). A Hypothalamic Midbrain Pathway Essential for Driving Maternal Behaviors. Neuron 98, 192–207 e110. 10.1016/j.neuron.2018.02.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Henschen CW, Palmiter RD, and Darvas M (2013). Restoration of dopamine signaling to the dorsal striatum is sufficient for aspects of active maternal behavior in female mice. Endocrinology 154, 4316–4327. 10.1210/en.2013-1257. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Numan M, Numan MJ, Pliakou N, Stolzenberg DS, Mullins OJ, Murphy JM, and Smith CD (2005). The effects of D1 or D2 dopamine receptor antagonism in the medial preoptic area, ventral pallidum, or nucleus accumbens on the maternal retrieval response and other aspects of maternal behavior in rats. Behav Neurosci 119, 1588–1604. 10.1037/0735-7044.119.6.1588. [DOI] [PubMed] [Google Scholar]
31.Numan M, Stolzenberg DS, Dellevigne AA, Correnti CM, and Numan MJ (2009). Temporary inactivation of ventral tegmental area neurons with either muscimol or baclofen reversibly disrupts maternal behavior in rats through different underlying mechanisms. Behav Neurosci 123, 740–751. 10.1037/a0016204. [DOI] [PubMed] [Google Scholar]
32.Afonso VM, King S, Chatterjee D, and Fleming AS (2009). Hormones that increase maternal responsiveness affect accumbal dopaminergic responses to pup- and food-stimuli in the female rat. Horm Behav 56, 11–23. 10.1016/j.yhbeh.2009.02.003. [DOI] [PubMed] [Google Scholar]
33.Afonso VM, Shams WM, Jin D, and Fleming AS (2013). Distal pup cues evoke dopamine responses in hormonally primed rats in the absence of pup experience or ongoing maternal behavior. J Neurosci 33, 2305–2312. 10.1523/JNEUROSCI.2081-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Champagne FA, Chretien P, Stevenson CW, Zhang TY, Gratton A, and Meaney MJ (2004). Variations in nucleus accumbens dopamine associated with individual differences in maternal behavior in the rat. J Neurosci 24, 4113–4123. 10.1523/JNEUROSCI.5322-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Hansen S, Bergvall AH, and Nyiredi S (1993). Interaction with pups enhances dopamine release in the ventral striatum of maternal rats: a microdialysis study. Pharmacol Biochem Behav 45, 673–676. 10.1016/0091-3057(93)90523-v. [DOI] [PubMed] [Google Scholar]
36.Rincon-Cortes M, and Grace AA (2020). Postpartum changes in affect-related behavior and VTA dopamine neuron activity in rats. Prog Neuropsychopharmacol Biol Psychiatry 97, 109768. 10.1016/j.pnpbp.2019.109768. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Shnitko TA, Mace KD, Sullivan KM, Martin WK, Andersen EH, Williams Avram SK, Johns JM, and Robinson DL (2017). Use of fast-scan cyclic voltammetry to assess phasic dopamine release in rat models of early postpartum maternal behavior and neglect. Behav Pharmacol 28, 648–660. 10.1097/FBP.0000000000000347. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Winokur SB, Lopes KL, Moparthi Y, and Pereira M (2019). Depression-related disturbances in rat maternal behaviour are associated with altered monoamine levels within mesocorticolimbic structures. J Neuroendocrinol 31, e12766. 10.1111/jne.12766. [DOI] [PubMed] [Google Scholar]
39.Dai B, Sun F, Tong X, Ding Y, Kuang A, Osakada T, Li Y, and Lin D (2022). Responses and functions of dopamine in nucleus accumbens core during social behaviors. Cell Rep 40, 111246. 10.1016/j.celrep.2022.111246. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Dana H, Sun Y, Mohar B, Hulse BK, Kerlin AM, Hasseman JP, Tsegaye G, Tsang A, Wong A, Patel R, et al. (2019). High-performance calcium sensors for imaging activity in neuronal populations and microcompartments. Nat Methods 16, 649–657. 10.1038/s41592-019-0435-6. [DOI] [PubMed] [Google Scholar]
41.Mathis A, Mamidanna P, Cury KM, Abe T, Murthy VN, Mathis MW, and Bethge M (2018). DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci 21, 1281–1289. 10.1038/s41593-018-0209-y. [DOI] [PubMed] [Google Scholar]
42.Gandelman R, Zarrow MX, and Denenberg VH ( 1971). Maternal behavior: differences between mother and virgin mice as a function of the testing procedure. Developmental psychobiology 3, 207–214. [DOI] [PubMed] [Google Scholar]
43.Mahn M, Gibor L, Patil P, Cohen-Kashi Malina K, Oring S, Printz Y, Levy R, Lampl I, and Yizhar O (2018). High-efficiency optogenetic silencing with soma-targeted anion-conducting channelrhodopsins. Nat Commun 9, 4125. 10.1038/s41467-018-06511-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Kohl J, Babayan BM, Rubinstein ND, Autry AE, Marin-Rodriguez B, Kapoor V, Miyamishi K, Zweifel LS, Luo L, Uchida N, and Dulac C (2018). Functional circuit architecture underlying parental behaviour. Nature 556, 326–331. 10.1038/s41586-018-0027-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Engelhard B, Finkelstein J, Cox J, Fleming W, Jang HJ, Ornelas S, Koay SA, Thiberge SY, Daw ND, Tank DW, and Witten IB (2019). Specialized coding of sensory, motor and cognitive variables in VTA dopamine neurons. Nature 570, 509–513. 10.1038/s41586-019-1261-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Mohebi A, Pettibone JR, Hamid AA, Wong JT, Vinson LT, Patriarchi T, Tian L, Kennedy RT, and Berke JD (2019). Dissociable dopamine dynamics for learning and motivation. Nature 570, 65–70. 10.1038/s41586-019-1235-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Lammel S, Lim BK, Ran C, Huang KW, Betley MJ, Tye KM, Deisseroth K, and Malenka RC (2012). Input-specific control of reward and aversion in the ventral tegmental area. Nature 491, 212–217. 10.1038/nature11527. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Kim HR, Malik AN, Mikhael JG, Bech P, Tsutsui-Kimura I, Sun F, Zhang Y, Li Y, Watabe-Uchida M, Gershman SJ, and Uchida N (2020). A Unified Framework for Dopamine Signals across Timescales. Cell 183, 1600–1616 e1625. 10.1016/j.cell.2020.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Lerner TN, Holloway AL, and Seiler JL (2021). Dopamine, Updated: Reward Prediction Error and Beyond. Curr Opin Neurobiol 67, 123–130. 10.1016/j.conb.2020.10.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Numan M, Bress JA, Ranker LR, Gary AJ, Denicola AL, Bettis JK, and Knapp SE (2010). The importance of the basolateral/basomedial amygdala for goal-directed maternal responses in postpartum rats. Behav Brain Res 214, 368–376. 10.1016/j.bbr.2010.06.006. [DOI] [PubMed] [Google Scholar]
51.Numan M, and Stolzenberg DS (2009). Medial preoptic area interactions with dopamine neural systems in the control of the onset and maintenance of maternal behavior in rats. Front Neuroendocrinol 30, 46–64. 10.1016/j.yfrne.2008.10.002. [DOI] [PubMed] [Google Scholar]
52.Carcea I, Caraballo NL, Marlin BJ, Ooyama R, Riceberg JS, Mendoza Navarro JM, Opendak M, Diaz VE, Schuster L, Alvarado Torres MI, et al. (2021). Oxytocin neurons enable social transmission of maternal behaviour. Nature 596, 553–557. 10.1038/s41586-021-03814-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Marlin BJ, Mitre M, D’Amour J A, Chao MV, and Froemke RC (2015). Oxytocin enables maternal behaviour by balancing cortical inhibition. Nature 520, 499–504. 10.1038/nature14402. [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Dvorkin R, and Shea SD (2022). Precise and pervasive phasic bursting in locus coeruleus during maternal behavior in mice. J Neurosci. 10.1523/JNEUROSCI.0938-21.2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Thomas SA, and Palmiter RD (1997). Impaired maternal behavior in mice lacking norepinephrine and epinephrine. Cell 91, 583–592. 10.1016/s0092-8674(00)80446-8. [DOI] [PubMed] [Google Scholar]
56.Gadagkar V, Puzerey PA, Chen R, Baird-Daniel E, Farhang AR, and Goldberg JH (2016). Dopamine neurons encode performance error in singing birds. Science 354, 1278–1282. 10.1126/science.aah6837. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Hisey E, Kearney MG, and Mooney R (2018). A common neural circuit mechanism for internally guided and externally reinforced forms of motor learning. Nat Neurosci 21, 589–597. 10.1038/s41593-018-0092-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Tanaka M, Sun F, Li Y, and Mooney R (2018). A mesocortical dopamine circuit enables the cultural transmission of vocal behaviour. Nature 563, 117–120. 10.1038/s41586-018-0636-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Willmore L, Cameron C, Yang J, Witten IB, and Falkner AL (2022). Behavioural and dopaminergic signatures of resilience. Nature 611, 124–132. 10.1038/s41586-022-05328-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Deaner RO, Khera AV, and Platt ML (2005). Monkeys pay per view: adaptive valuation of social images by rhesus macaques. Current biology : CB 15, 543–548. 10.1016/j.cub.2005.01.044. [DOI] [PubMed] [Google Scholar]
61.Hayden BY, Parikh PC, Deaner RO, and Platt ML (2007). Economic principles motivating social attention in humans. Proc Biol Sci 274, 1751–1756. 10.1098/rspb.2007.0368. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Krishnan K, Lau BY, Ewall G, Huang ZJ, and Shea SD (2017). MECP2 regulates cortical plasticity underlying a learned behaviour in adult female mice. Nat Commun 8, 14077. 10.1038/ncomms14077. [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Lopes G, Bonacchi N, Frazao J, Neto JP, Atallah BV, Soares S, Moreira L, Matias S, Itskov PM, Correia PA, et al. (2015). Bonsai: an event-based framework for processing and controlling data streams. Front Neuroinform 9, 7. 10.3389/fninf.2015.00007. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS1856976-supplement-2.pdf^{(8.4MB, pdf)}

Download video file^{(45.2MB, mov)}

Download video file^{(15.3MB, mp4)}

Data Availability Statement

All data reported in this study are available from the lead contact upon request.

All original code has been deposited at Zenodo and is publicly available now. The DOI is listed in the key resources table.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken anti-TH	Abcam	Cat #76442; RRID: AB_1524535
Rabbit anti-GFP	Invitrogen	Cat# A6455; RRID: AB_221570
Goat anti-chicken Alexa Fluor 594	Invitrogen	Cat# A11042; RRID: AB_2534099
Goat anti-chicken Alexa Fluor 488	Invitrogen	Cat# A11039; RRID: AB_2534096
Goat anti-rabbit Alexa Fluor 488	Invitrogen	Cat# A11008; RRID: AB_143165
Bacterial and virus strains
AAV-syn-FLEX-jGCaMP7f-WPRE	Addgene	Cat# 104492-AAV9
AAV-hSyn1-SIO-stGtACR2-FusionRed	Addgene	Cat# 105677-AAV1
AAV-synP-DIO-EGFP-WPRE-hGH	Addgene	Cat# 100043-AAV9
Chemicals, peptides, and recombinant proteins
DirectPCR	Viagen	Cat# 402-E
Proteinase K	Invitrogen	Cat# 2036707
Euthasol	Virbac	Cat# 200-071
4% paraformaldehyde	FD Neurotechnologies	Cat# PF101
OCT compound	Sakura	Cat# 4583
Normal goat serum	Vector Laboratories	Cat# S-1000-20
Triton X-100	Sigma	Cat# T8787
Ketamine	Ketaset	Cat# NDC 54771-2013
Xylazine	AnaSed	Cat# NDC 59399-110-20
Meloxicam	Metacam	Cat# NDC 0010-6013-01
Baytril	Bayer	Cat# 08713254-186599
Metabond	C&B	Cat# S380
Critical commercial assay
GoTaq Green Master Mix	Promega	M7123
Experimental models: Organisms/strains
DAT-Cre (Slc6a3tm1.1(cre)Bkmn)	The Jackson Laboratory	Cat# 006302
CBA/CaJ	The Jackson Laboratory	Cat# 000654
Software and algorithms
Zen	Zeiss	https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html
MATLAB R2019b	MatWorks	https://www.mathworks.com/products/matlab.html
Spike2	Cambridge Electronic Design	https://ced.co.uk/products/spkovin
Custom analysis code		http://doi.org/10.5281/zenodo.7377063
Other
Microtome	Leica	SM 2010R
Confocal microscope	Zeiss	LSM 710
Optic fibers for fiber photometry	Thorlabs	Cat# CFMLC12U
Dual fiber cannulae for optogenetics	Doric lenses	Cat# B280-2013_6
Industrial camera BlackFly S	Teledyne FLIR	Cat# FL3-U3-13Y3M-C
Step motor	McMaster-Carr	Cat# 6627T33
Step motor driver	McMaster-Carr	Cat# 6627T41
Arduino Uno R3	Arduino	Cat# A000066
Pulse Pal v2	Sanworks	Cat# 1102
Amplifier	Brownlee Precision	Cat# 410
TDT electrostatic speaker and driver	TDT	Cat# ES1 and ED1
Patch cord for fiber photometry	Doric Lenses	Cat# P99414-01
Mating sleeves	Thorlabs	Cat# ADAL1
470nm and 565nm LED light sources	Thorlabs	Cat# M470F3 and M565F3
LED driver	Thorlabs	Cat# LEDD1B
Photoreceivers	Newport	Cat# 2151
NIDAQ boards	National Instruments	Cat# USB-6211 and USB-6001
Rotary joint for optogenetics	Doric lenses	Cat# FRJ_1×2i_FC-2FC
473nm laser and laser driver	OEM Laser Systems	Cat# PSU III LED
1.0 MΩ tungsten microelectrode	Microprobe	Cat# WE30031.0A3
AC differential amplifier (model 1800)	A–M Systems	Cat# 700000

Open in a new tab

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

[R1] 1.Krach S, Paulus FM, Bodden M, and Kircher T (2010). The rewarding nature of social interactions. Front Behav Neurosci 4, 22. 10.3389/fnbeh.2010.00022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] 2.Dolen G, Darvishzadeh A, Huang KW, and Malenka RC (2013). Social reward requires coordinated activity of nucleus accumbens oxytocin and serotonin. Nature 501, 179–184. 10.1038/nature12518. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Hung LW, Neuner S, Polepalli JS, Beier KT, Wright M, Walsh JJ, Lewis EM, Luo L, Deisseroth K, Dolen G, and Malenka RC (2017). Gating of social reward by oxytocin in the ventral tegmental area. Science 357, 1406–1411. 10.1126/science.aan4994. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.McHenry JA, Otis JM, Rossi MA, Robinson JE, Kosyk O, Miller NW, McElligott ZA, Budygin EA, Rubinow DR, and Stuber GD (2017). Hormonal gain control of a medial preoptic area social reward circuit. Nat Neurosci 20, 449–458. 10.1038/nn.4487. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Nardou R, Lewis EM, Rothhaas R, Xu R, Yang A, Boyden E, and Dolen G (2019). Oxytocin-dependent reopening of a social reward learning critical period with MDMA. Nature 569, 116–120. 10.1038/s41586-019-1075-9. [DOI] [PubMed] [Google Scholar]

[R6] 6.Resendez SL, Namboodiri VMK, Otis JM, Eckman LEH, Rodriguez-Romaguera J, Ung RL, Basiri ML, Kosyk O, Rossi MA, Dichter GS, and Stuber GD (2020). Social Stimuli Induce Activation of Oxytocin Neurons Within the Paraventricular Nucleus of the Hypothalamus to Promote Social Behavior in Male Mice. J Neurosci 40, 2282–2295. 10.1523/JNEUROSCI.1515-18.2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Solie C, Girard B, Righetti B, Tapparel M, and Bellone C (2022). VTA dopamine neuron activity encodes social interaction and promotes reinforcement learning through social prediction error. Nat Neurosci 25, 86–97. 10.1038/s41593-021-00972-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Montague PR, Dayan P, and Sejnowski TJ (1996). A framework for mesencephalic dopamine systems based on predictive Hebbian learning. J Neurosci 16, 1936–1947. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Sutton RS, and Barto AG (1998). Reinforcement Learning: An Introduction (MIT Press). [Google Scholar]

[R10] 10.Schultz W, Dayan P, and Montague PR (1997). A neural substrate of prediction and reward. Science 275, 1593–1599. 10.1126/science.275.5306.1593. [DOI] [PubMed] [Google Scholar]

[R11] 11.Watabe-Uchida M, Eshel N, and Uchida N (2017). Neural Circuitry of Reward Prediction Error. Annu Rev Neurosci 40, 373–394. 10.1146/annurev-neuro-072116-031109. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Berke JD (2018). What does dopamine mean? Nat Neurosci 21, 787–793. 10.1038/s41593-018-0152-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Hamid AA, Pettibone JR, Mabrouk OS, Hetrick VL, Schmidt R, Vander Weele CM, Kennedy RT, Aragona BJ, and Berke JD (2016). Mesolimbic dopamine signals the value of work. Nat Neurosci 19, 117–126. 10.1038/nn.4173. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Barter JW, Li S, Lu D, Bartholomew RA, Rossi MA, Shoemaker CT, Salas-Meza D, Gaidis E, and Yin HH (2015). Beyond reward prediction errors: the role of dopamine in movement kinematics. Front Integr Neurosci 9, 39. 10.3389/fnint.2015.00039. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Bova A, Gaidica M, Hurst A, Iwai Y, Hunter J, and Leventhal DK (2020). Precisely timed dopamine signals establish distinct kinematic representations of skilled movements. Elife 9. 10.7554/eLife.61591. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.da Silva JA, Tecuapetla F, Paixao V, and Costa RM (2018). Dopamine neuron activity before action initiation gates and invigorates future movements. Nature 554, 244–248. 10.1038/nature25457. [DOI] [PubMed] [Google Scholar]

[R17] 17.Hughes RN, Bakhurin KI, Petter EA, Watson GDR, Kim N, Friedman AD, and Yin HH (2020). Ventral Tegmental Dopamine Neurons Control the Impulse Vector during Motivated Behavior. Curr Biol 30, 2681–2694 e2685. 10.1016/j.cub.2020.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Ehret G, Koch M, Haack B, and Markl H (1987). Sex and parental experience determine the onset of an instinctive behavior in mice. Naturwissenschaften 74, 47. [DOI] [PubMed] [Google Scholar]

[R19] 19.Stolzenberg DS, and Champagne FA (2016). Hormonal and non-hormonal bases of maternal behavior: The role of experience and epigenetic mechanisms. Horm Behav 77, 204–210. 10.1016/j.yhbeh.2015.07.005. [DOI] [PubMed] [Google Scholar]

[R20] 20.Wilsoncroft WE ( 1969). Babies by bar press: maternal behavior in the rat. Behavioral research methods 1, 229–230. [Google Scholar]

[R21] 21.Fleming AS, Korsmit M, and Deller M (1994). Rat pups are potent reinforcers to the maternal animal: effects of experience, parity, hormones, and dopamine function. Psychobiology 22, 44–53. [Google Scholar]

[R22] 22.Hauser H, and Gandelman R (1985). Lever pressing for pups: evidence for hormonal influence upon maternal behavior of mice. Horm Behav 19, 454–468. [DOI] [PubMed] [Google Scholar]

[R23] 23.Dunlap AG, Besosa C, Pascual LM, Chong KK, Walum H, Kacsoh DB, Tankeu BB, Lu K, and Liu RC (2020). Becoming a better parent: Mice learn sounds that improve a stereotyped maternal behavior. Horm Behav 124, 104779. 10.1016/j.yhbeh.2020.104779. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Mattson BJ, Williams S, Rosenblatt JS, and Morrell JI (2001). Comparison of two positive reinforcing stimuli: pups and cocaine throughout the postpartum period. Behav Neurosci 115, 683–694. [DOI] [PubMed] [Google Scholar]

[R25] 25.Curry T, Egeto P, Wang H, Podnos A, Wasserman D, and Yeomans J (2013). Dopamine receptor D2 deficiency reduces mouse pup ultrasonic vocalizations and maternal responsiveness. Genes Brain Behav 12, 397–404. 10.1111/gbb.12037. [DOI] [PubMed] [Google Scholar]

[R26] 26.Seip KM, and Morrell JI (2009). Transient inactivation of the ventral tegmental area selectively disrupts the expression of conditioned place preference for pup-but not cocaine-paired contexts. Behav Neurosci 123, 1325–1338. 10.1037/a0017666. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Stolzenberg DS, Zhang KY, Luskin K, Ranker L, Bress J, and Numan M (2010). Dopamine D(1) receptor activation of adenylyl cyclase, not phospholipase C, in the nucleus accumbens promotes maternal behavior onset in rats. Horm Behav 57, 96–104. 10.1016/j.yhbeh.2009.09.014. [DOI] [PubMed] [Google Scholar]

[R28] 28.Fang YY, Yamaguchi T, Song SC, Tritsch NX, and Lin D (2018). A Hypothalamic Midbrain Pathway Essential for Driving Maternal Behaviors. Neuron 98, 192–207 e110. 10.1016/j.neuron.2018.02.019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Henschen CW, Palmiter RD, and Darvas M (2013). Restoration of dopamine signaling to the dorsal striatum is sufficient for aspects of active maternal behavior in female mice. Endocrinology 154, 4316–4327. 10.1210/en.2013-1257. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Numan M, Numan MJ, Pliakou N, Stolzenberg DS, Mullins OJ, Murphy JM, and Smith CD (2005). The effects of D1 or D2 dopamine receptor antagonism in the medial preoptic area, ventral pallidum, or nucleus accumbens on the maternal retrieval response and other aspects of maternal behavior in rats. Behav Neurosci 119, 1588–1604. 10.1037/0735-7044.119.6.1588. [DOI] [PubMed] [Google Scholar]

[R31] 31.Numan M, Stolzenberg DS, Dellevigne AA, Correnti CM, and Numan MJ (2009). Temporary inactivation of ventral tegmental area neurons with either muscimol or baclofen reversibly disrupts maternal behavior in rats through different underlying mechanisms. Behav Neurosci 123, 740–751. 10.1037/a0016204. [DOI] [PubMed] [Google Scholar]

[R32] 32.Afonso VM, King S, Chatterjee D, and Fleming AS (2009). Hormones that increase maternal responsiveness affect accumbal dopaminergic responses to pup- and food-stimuli in the female rat. Horm Behav 56, 11–23. 10.1016/j.yhbeh.2009.02.003. [DOI] [PubMed] [Google Scholar]

[R33] 33.Afonso VM, Shams WM, Jin D, and Fleming AS (2013). Distal pup cues evoke dopamine responses in hormonally primed rats in the absence of pup experience or ongoing maternal behavior. J Neurosci 33, 2305–2312. 10.1523/JNEUROSCI.2081-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Champagne FA, Chretien P, Stevenson CW, Zhang TY, Gratton A, and Meaney MJ (2004). Variations in nucleus accumbens dopamine associated with individual differences in maternal behavior in the rat. J Neurosci 24, 4113–4123. 10.1523/JNEUROSCI.5322-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Hansen S, Bergvall AH, and Nyiredi S (1993). Interaction with pups enhances dopamine release in the ventral striatum of maternal rats: a microdialysis study. Pharmacol Biochem Behav 45, 673–676. 10.1016/0091-3057(93)90523-v. [DOI] [PubMed] [Google Scholar]

[R36] 36.Rincon-Cortes M, and Grace AA (2020). Postpartum changes in affect-related behavior and VTA dopamine neuron activity in rats. Prog Neuropsychopharmacol Biol Psychiatry 97, 109768. 10.1016/j.pnpbp.2019.109768. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] 37.Shnitko TA, Mace KD, Sullivan KM, Martin WK, Andersen EH, Williams Avram SK, Johns JM, and Robinson DL (2017). Use of fast-scan cyclic voltammetry to assess phasic dopamine release in rat models of early postpartum maternal behavior and neglect. Behav Pharmacol 28, 648–660. 10.1097/FBP.0000000000000347. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Winokur SB, Lopes KL, Moparthi Y, and Pereira M (2019). Depression-related disturbances in rat maternal behaviour are associated with altered monoamine levels within mesocorticolimbic structures. J Neuroendocrinol 31, e12766. 10.1111/jne.12766. [DOI] [PubMed] [Google Scholar]

[R39] 39.Dai B, Sun F, Tong X, Ding Y, Kuang A, Osakada T, Li Y, and Lin D (2022). Responses and functions of dopamine in nucleus accumbens core during social behaviors. Cell Rep 40, 111246. 10.1016/j.celrep.2022.111246. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] 40.Dana H, Sun Y, Mohar B, Hulse BK, Kerlin AM, Hasseman JP, Tsegaye G, Tsang A, Wong A, Patel R, et al. (2019). High-performance calcium sensors for imaging activity in neuronal populations and microcompartments. Nat Methods 16, 649–657. 10.1038/s41592-019-0435-6. [DOI] [PubMed] [Google Scholar]

[R41] 41.Mathis A, Mamidanna P, Cury KM, Abe T, Murthy VN, Mathis MW, and Bethge M (2018). DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat Neurosci 21, 1281–1289. 10.1038/s41593-018-0209-y. [DOI] [PubMed] [Google Scholar]

[R42] 42.Gandelman R, Zarrow MX, and Denenberg VH ( 1971). Maternal behavior: differences between mother and virgin mice as a function of the testing procedure. Developmental psychobiology 3, 207–214. [DOI] [PubMed] [Google Scholar]

[R43] 43.Mahn M, Gibor L, Patil P, Cohen-Kashi Malina K, Oring S, Printz Y, Levy R, Lampl I, and Yizhar O (2018). High-efficiency optogenetic silencing with soma-targeted anion-conducting channelrhodopsins. Nat Commun 9, 4125. 10.1038/s41467-018-06511-8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Kohl J, Babayan BM, Rubinstein ND, Autry AE, Marin-Rodriguez B, Kapoor V, Miyamishi K, Zweifel LS, Luo L, Uchida N, and Dulac C (2018). Functional circuit architecture underlying parental behaviour. Nature 556, 326–331. 10.1038/s41586-018-0027-0. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Engelhard B, Finkelstein J, Cox J, Fleming W, Jang HJ, Ornelas S, Koay SA, Thiberge SY, Daw ND, Tank DW, and Witten IB (2019). Specialized coding of sensory, motor and cognitive variables in VTA dopamine neurons. Nature 570, 509–513. 10.1038/s41586-019-1261-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Mohebi A, Pettibone JR, Hamid AA, Wong JT, Vinson LT, Patriarchi T, Tian L, Kennedy RT, and Berke JD (2019). Dissociable dopamine dynamics for learning and motivation. Nature 570, 65–70. 10.1038/s41586-019-1235-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Lammel S, Lim BK, Ran C, Huang KW, Betley MJ, Tye KM, Deisseroth K, and Malenka RC (2012). Input-specific control of reward and aversion in the ventral tegmental area. Nature 491, 212–217. 10.1038/nature11527. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Kim HR, Malik AN, Mikhael JG, Bech P, Tsutsui-Kimura I, Sun F, Zhang Y, Li Y, Watabe-Uchida M, Gershman SJ, and Uchida N (2020). A Unified Framework for Dopamine Signals across Timescales. Cell 183, 1600–1616 e1625. 10.1016/j.cell.2020.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Lerner TN, Holloway AL, and Seiler JL (2021). Dopamine, Updated: Reward Prediction Error and Beyond. Curr Opin Neurobiol 67, 123–130. 10.1016/j.conb.2020.10.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Numan M, Bress JA, Ranker LR, Gary AJ, Denicola AL, Bettis JK, and Knapp SE (2010). The importance of the basolateral/basomedial amygdala for goal-directed maternal responses in postpartum rats. Behav Brain Res 214, 368–376. 10.1016/j.bbr.2010.06.006. [DOI] [PubMed] [Google Scholar]

[R51] 51.Numan M, and Stolzenberg DS (2009). Medial preoptic area interactions with dopamine neural systems in the control of the onset and maintenance of maternal behavior in rats. Front Neuroendocrinol 30, 46–64. 10.1016/j.yfrne.2008.10.002. [DOI] [PubMed] [Google Scholar]

[R52] 52.Carcea I, Caraballo NL, Marlin BJ, Ooyama R, Riceberg JS, Mendoza Navarro JM, Opendak M, Diaz VE, Schuster L, Alvarado Torres MI, et al. (2021). Oxytocin neurons enable social transmission of maternal behaviour. Nature 596, 553–557. 10.1038/s41586-021-03814-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R53] 53.Marlin BJ, Mitre M, D’Amour J A, Chao MV, and Froemke RC (2015). Oxytocin enables maternal behaviour by balancing cortical inhibition. Nature 520, 499–504. 10.1038/nature14402. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] 54.Dvorkin R, and Shea SD (2022). Precise and pervasive phasic bursting in locus coeruleus during maternal behavior in mice. J Neurosci. 10.1523/JNEUROSCI.0938-21.2022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] 55.Thomas SA, and Palmiter RD (1997). Impaired maternal behavior in mice lacking norepinephrine and epinephrine. Cell 91, 583–592. 10.1016/s0092-8674(00)80446-8. [DOI] [PubMed] [Google Scholar]

[R56] 56.Gadagkar V, Puzerey PA, Chen R, Baird-Daniel E, Farhang AR, and Goldberg JH (2016). Dopamine neurons encode performance error in singing birds. Science 354, 1278–1282. 10.1126/science.aah6837. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Hisey E, Kearney MG, and Mooney R (2018). A common neural circuit mechanism for internally guided and externally reinforced forms of motor learning. Nat Neurosci 21, 589–597. 10.1038/s41593-018-0092-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Tanaka M, Sun F, Li Y, and Mooney R (2018). A mesocortical dopamine circuit enables the cultural transmission of vocal behaviour. Nature 563, 117–120. 10.1038/s41586-018-0636-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] 59.Willmore L, Cameron C, Yang J, Witten IB, and Falkner AL (2022). Behavioural and dopaminergic signatures of resilience. Nature 611, 124–132. 10.1038/s41586-022-05328-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R60] 60.Deaner RO, Khera AV, and Platt ML (2005). Monkeys pay per view: adaptive valuation of social images by rhesus macaques. Current biology : CB 15, 543–548. 10.1016/j.cub.2005.01.044. [DOI] [PubMed] [Google Scholar]

[R61] 61.Hayden BY, Parikh PC, Deaner RO, and Platt ML (2007). Economic principles motivating social attention in humans. Proc Biol Sci 274, 1751–1756. 10.1098/rspb.2007.0368. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Krishnan K, Lau BY, Ewall G, Huang ZJ, and Shea SD (2017). MECP2 regulates cortical plasticity underlying a learned behaviour in adult female mice. Nat Commun 8, 14077. 10.1038/ncomms14077. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R63] 63.Lopes G, Bonacchi N, Frazao J, Neto JP, Atallah BV, Soares S, Moreira L, Matias S, Itskov PM, Correia PA, et al. (2015). Bonsai: an event-based framework for processing and controlling data streams. Front Neuroinform 9, 7. 10.3389/fninf.2015.00007. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

A dopaminergic reward prediction error signal shapes maternal behavior in mice

Yunyao Xie

Longwen Huang

Alberto Corona

Alexa H Pagliaro

Stephen D Shea

SUMMARY

eTOC blurb

INTRODUCTION

Figure 1: A brief phasic burst in the population activity of VTA DA neurons accompanies each pup retrieval event.

RESULTS

VTA DA neuron signals during retrieval are inversely correlated with performance

Figure 2: Population activity of VTA DA neurons during maternal retrieval is inversely related to retrieval performance by session.

Figure 3: Retrieval behavior is updated from trial to trial based on VTA DA neuron signals.

Retrieval performance is influenced by the history of DA neuron activity

VTA DA neurons signal reward prediction error for pup contact

Figure 4: VTA DA neuron responses shift from pup contact to door opening over the course of retrieval learning.

Figure 5: Retrieval activity is modulated by expectation of pup contact.

Optogenetic inhibition of VTA DA neuron activity impairs pup retrieval

Figure 6: Optogenetic inhibition of VTA DA neuron activity results in slower pup retrieval.

Activity of VTA DA neurons is required for reinforcement but not execution of pup retrieval

Figure 7: VTA DA neuron activity at pup contact gradually reinforces pup retrieval.

DISCUSSION

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Lead contact

Materials availability

Data and code availability

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Genotyping

Histology

Imaging and image analysis

Viruses

Stereotaxic surgery

Behavioral annotation and tracking

Fiber photometry

Closed-loop optogenetics

In vivo extracellular electrophysiology

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics

Supplementary Material

Highlights.

ACKNOWLEDGMENTS

Footnotes

REFERENCES

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases