Impaired emotion recognition in Cntnap2-deficient mice is associated with hyper-synchronous prefrontal cortex neuronal activity

Individuals diagnosed with autism spectrum disorder (ASD) show difficulty in recognizing emotions in others, a process termed emotion recognition. While human fMRI studies linked multiple brain areas to emotion recognition, the specific mechanisms underlying impaired emotion recognition in ASD are not clear, partially due to the lack of appropriate tests in animal models. Here, we employed an emotional state preference (ESP) task to show that Cntnap2-knockout (KO) mice, an established ASD model, do not distinguish between conspecifics according to their emotional state. We assessed brain-wide local-field potential (LFP) signals during various social behavior tasks and found that Cntnap2-KO mice exhibited higher LFP theta and gamma rhythmicity than did C57BL/6J mice, even at rest. Specifically, Cntnap2-KO mice showed increased theta coherence, especially between the prelimbic cortex (PrL) and the hypothalamic paraventricular nucleus, during social behavior. Moreover, we observed significantly increased Granger causality of theta rhythmicity between these two brain areas, across all types of social behavior. Finally, optogenetic stimulation of PrL pyramidal neurons in C57BL/6J mice impaired their social discrimination abilities, including ESP behavior. Together, these results suggest that increased activity of PrL pyramidal neurons and their augmented synchronization with specific brain regions are involved in the impaired emotion recognition exhibited by Cntnap2-KO mice.


Introduction
Social cognition involves the perception and interpretation of social cues transmitted between individuals, processes crucial for the appropriate adaptation of a subject to its social environment 1,2 .The ability to recognize the emotional state of other individuals, termed emotion recognition, is crucial for a wide range of prosocial behaviors, such as emotion contagion, empathy and helping behavior 3,4 and is known to be impaired in Individuals diagnosed with autism spectrum disorder (ASD) [5][6][7] .While fMRI studies have supplied considerable information regarding active brain areas during tasks of emotion recognition [6][7][8][9][10] , including frontal lobe and amygdalar regions 11,12 , specific deficits in brain activity and neuronal network dynamics that lead to impaired emotion recognition in ASD remain elusive.One valuable tool for deciphering such deficits is animal models of ASD, which allow for invasive monitoring and manipulation of brain neural activity during social behavior.Yet, reliable behavioral tasks to assess emotion recognition in animal models were lacking until only recently.
Of late, we and others have demonstrated that mice can discriminate between conspecifics according to the conspecific's emotional state, thus providing a tool for assessing emotion recognition in murine models of ASD [13][14][15] .This observation led to the development of a new behavioral task, termed by us emotional state preference (ESP).
This task is based on the ability to discriminate between two stimulus animals simultaneously presented to a subject mouse, one of which was emotionally aroused by a given manipulation.While Ferretti et al. showed that C57BL/6J mice preferred to investigate stressed and relieved conspecifics more than neutral stimulus animals 13 , we demonstrated that C57BL/6J mice preferred to investigate a stimulus animal socially isolated for seven days over a group-housed stimulus animal 14 .
We previously demonstrated that mice expressing the A350V-encoding mutation in the Iqsec2 gene, a mutation associated with ASD in humans 16 , showed a specific deficit in the ESP task 14 .Here, we sought to further validate this observation in another ASD murine model, so as to assess the generality of this deficit.One of the most established murine models of ASD is the Cntanp2-knockout (KO) mouse line, which does not express a functional copy of the Contactin-associated protein 2 gene 17 .Previous studies showed that such mice exhibit a reduced tendency for social interaction and that this can be reversed by inhibiting the excitability of medial prefrontal cortex (mPFC) pyramidal neurons or by activating inhibitory GABAergic interneurons in this region 18 .Natably, recognition of stress and relieved states by C57BL/6J mice was found to be dependent upon somatostatin-expressing GABA-ergic inhibitory interneurons in the mPFC 15 .
However, emotion recognition by Cntnap2-KO mice has yet to be examined.
Here, we addressed this issue by employing the ESP paradigm to examine emotion recognition in Cntnap2-KO mice.We found that these mice do not prefer to investigate aroused over neutral conspecifics.Surprisingly, we found that even wild-type (WT) offspring of Cntnap2 -/+ mice were impaired in this behavior, suggesting that the etiology of this impairment involves not only a subject's genotype, but also the genotype of its parents and/or littermates.We further simultaneously recorded local field potential (LFP) signals from multiple brain areas linked to social behavior and found hyperactive theta and gamma rhythmicity in the brains of Cntnap2-KO mice, as compared to C57BL/6J mice.Specifically, synchronization between the mPFC and several hypothalamic and amygdalar areas was consistently modified in Cntnap2-KO mice across multiple social discrimination tasks.Finally, using optogenetic stimulation, we demonstrated that stimulating mPFC pyramidal neurons at either theta or gamma frequency impaired the ability of C57BL/6J mice to discriminate between various types of conspecifics and specifically, between emotionally-aroused and neutral stimuli, similarly to Cntnap2-KO mice.These results suggest that the modified mPFC activity exhibited by Cntnap2-KO mice does not merely cause social avoidance, as previously suggested, but rather interferes with social recognition and discrimination, thus creating a complex deficit which seems to be highly related to ASD.

Cntnap2-KO and WT littermates exhibit impaired emotion recognition
Since patients diagnosed with ASD are known to exhibit impaired emotion recognition 3- 5 , we first examined whether Cntnap2-KO mice are impaired in terms of ESP using two variations of the test that assesses this trait.Specifically, in the stress-state preference (SSP) task, subjects simultaneously encountered a stressed stimulus animal and a neutral animal, whereas in the isolation-state preference (ISP) task, subjects encountered a socially isolated stimulus animal and a group-housed animal.Each task consisted of a 5 min encounter session, which followed a 5 min baseline period during which time no stimulus was introduced into the arena.As a control, we conducted a social preference (SP) task in which the subjects encountered a novel animal stimulus and an object (Fig. 1A).In this task, both WT and Cntnap2 -/-(Cntnap2-KO) male littermates showed similar SP behavior, namely, investigating the stimulus animal for significantly more time than the object (Fig. 1D-E).Surprisingly, when conducting the SSP (Fig. 1B) and ISP (Fig. 1C) tasks with these same two genotypes, we found that both WT and KO mice did not discriminate between the two stimulus animals in either task (Fig. 1F-G, H-I).Since previous studies showed that behavioral deficits in ASD genetic models are sometimes exhibited by WT animals raised with mutant littermates by mutant parents 19,20 , we generated "pure WT" mice by breeding two WT parents 21 .We found that pure WT mice perform normally in the SSP and ISP tasks (Fig. 1J-K).These results show that Cntnap2-KO mice are impaired in terms of their ESP behavior, although the etiology of this impairment involves not only the subject's genotype but also the genotypes of its parents and/or littermates.As pure WT animals cannot be littermates of KO animals and since the genetic background of the KO mice is that of the C57BL/6J (C57) mouse strain, we continued this study by comparing brain activity of KO and C57 mice during social behavior.

J.
As in F, for pure WT animals produced by breeding two WT littermates (see the breeding scheme in the inset on the left) (t (18) = 2.403, P = 0.0273).K.As in J, for the ISP task (t (20) = 2.987, P = 0.0073).
L. As in K, using C57BL/6J female mice (t (20) = 1.429,P = 0.1658).See the breeding scheme in the inset on the left.M. As in K, using pure WT female mice (t (19) = 0.836, P = 0.4141).See the breeding scheme in the inset on the left.Accordingly, we employed four behavioral tasks, namely, the SP and the ISP tasks, a sex preference (SxP) task in which the subjects encountered a male and a female stimulus animal, and a free social interaction (FSI) task involving a same-sex, age-matched novel stimulus animal.For these experiments, we only used male subjects, as both C57 and pure WT female mice did not show a preference in the ISP task, which seems to be sexspecific (Fig. 1L-M).A multi-electrode array was implanted in the brain of each subject as previously described 22 and behavioral experiments were conducted 3 days later (see timeline in Fig. 2A).Each of the recorded subjects (11 C57 and 15 KO mice) performed up to three sessions of each task, as described in Fig. 2A, and the results from each session were separately analyzed.
We found that both genotypes (C57 and KO) showed a significant preference to investigate the social stimulus over the object in the SP task (Fig. 2B-C).However, while C57 mice also showed a preference to investigate the isolated stimulus animal in the ISP task (Fig. 2D) and the female stimulus animal in the SxP task (Fig. 2F), KO mice did not show any preferences in the two tasks (Fig. 2E, G).Interestingly, while there was no significant difference between the two genotypes in terms of the mean distance traveled in the arena during the encounter period of any task (Fig. 2I, K, M), KO mice displayed significantly longer total investigation times (of both stimuli together) in the ISP and SxP tasks (Fig. 2H, J, L).This suggests that KO mice are generally more exploratory than are C57 mice.In the last task (i.e., the FSI task), each subject was exposed to a novel stimulus animal for five minutes in the empty arena, and the behavior of the subject was tracked using DeepLabCut (DLC) software 23 , and analyzed by SimBA behavioral segmentation analysis (Fig. 2N and Fig. S1) 24 .We found no differences between the two genotypes in any of the behavioral variables analyzed (Fig. 2O-R), other than the number of "sitting idle" events (Fig. S1E).Thus, we identified deficits in the social behavior of KO mice, specifically in the ISP and SxP tasks.

Cntnap2-KO mice display augmented theta and gamma rhythmicity
To assess brain activity during behavior, we recorded local field potential (LFP) signals from microelectrode array-implanted mice while they performed various behavioral tasks (Fig. 3A).The location of each electrode tip was verified post-mortem in each mouse (Fig. S1K) and only regions where an adequate sample size was available for both C57 and KO mice were considered.Of all recorded brain areas (nine common to C57 and KO mice), we analyzed signals from seven social behavior-associated brain regions (see Methods, supp.Table 1).These brain areas addressed included the anterodorsal part of the medial amygdala (MeAD), the nucleus accumbens core (AsbC) and shell (AcbSh), the prelimbic (PrL) and infralimbic (IL) cortices, the hypothalamic paraventricular nucleus (PVN) and the lateral septum (LS) (Fig. 3B).
We first analyzed the mean power of theta and gamma rhythmicity of LFP signals recorded during the 5 min-long baseline period, during which time there were no stimuli in the arena, across all the brain regions listed above (Fig. 3C-D).We measured higher theta power in KO mice, as compared to C57 mice, in all four tasks, with this difference being statistically significant in all cases (Fig. 3E, G, I, K).Gamma power showed a similar trend, albeit without statistical significance (Fig 3F , H, J, L).These differences between the genotypes were observed even when we considered only the first session for each mouse, thus excluding the possibility that the changes were associated with expectation of a social encounter (Fig. S1L-M).The results suggest that KO mice exhibit a higher level of a certain internal state even before the beginning of the behavioral task.
We next analyzed the mean change (∆) in theta and gamma powers recorded during the 5 min-long encounter, as compared to baseline power values (Fig. 3D).In all cases, LFP power during the encounter was higher than at baseline, as reflected by positive ∆power values in both the theta and gamma bands (Fig. 3O-V).However, the KO mice exhibited significantly higher increases in theta rhythmicity in all three social discrimination tasks (Fig. 3M, O, Q), while the gamma power change was only significantly higher in the SxP task (Fig. 3N, P, R).In contrast to the discrimination tasks, we did not find a significant difference in power change between genotypes during the FSI task (Fig. 3S-T).Overall, these results suggest a higher internal state, reflected by stronger theta rhythmicity, in KO mice than in C57 mice, which was further enhanced during the encounter stage of the various social discrimination tasks.

M-T.
As in E-L, for ∆power during the encounter, as compared to baseline.M: t ( 12

Cntnap2-KO mice exhibit hyper-synchronized brain activity during social interaction
While the power of LFP rhythmicity may reflect the internal state of the animals 25,26 , coherence between brain regions is thought to reflect functional connectivity between them 27 .When analyzing the coherence of LFP rhythmicity across all pairs of brain regions considered, we found that the mean coherence did not differ between tasks and genotypes during the baseline period for both theta and gamma rhythms (Fig. 4A, C).
These results suggest that the mean coherence did not reflect the initial internal state that affected the mean LFP power (Fig. 3E-L).However, the normalized change in theta coherence during the encounter showed a significant difference (after correcting for multiple comparisons) between genotypes, with KO animals showing higher coherence changes in all tasks, other than the SP task (Fig. 4B).No significant differences were observed for the change in gamma coherence (Fig. 4D).Thus, theta rhythmicity during the various tasks seems to be hyper-synchronized across the recorded brain regions in KO mice.Notably, the coherence between the PrL and PVN seemed to be especially high in KO animals, as compared to C57 mice, for both theta and gamma rhythmicity (Fig. 4B, D, filled circles; see also heat-maps in Fig. 4E-L).We, therefore, analyzed the coherence between this pair of brain regions in each task separately and found a significantly higher encounter-induced change in theta coherence in KO mice in all cases (Fig. 4M-P).At the same time, changes in gamma coherence only showed a significant difference in the SP and SxP tasks (Fig. 4Q, S).These results suggest higher synchronization in theta rhythmicity of KO mice during social behavior, especially between the PrL and PVN.O.As in M, for SxP sessions (U = 0, P = 0.0006).

The prelimbic cortex in Cntnap2-KO mice shows modified synchronization during social interaction
To further explore the possibility that KO mice exhibit modified synchronization between brain region, we analyzed Granger's causality (GC), a measure which assesses predictability of the rhythmicity in one region according to the rhythmicity in another region, for a given pair of brain regions.The GC was separately calculated for each direction of the possible interaction between each pair of brain regions considered in this study.Interestingly, the GC level in the PrL to PVN direction in the theta band was consistently higher in KO mice than in C57 animals, with the difference being statistically significant (after correcting for multiple comparisons) across all tasks, other than the SxP task (Fig. 5 A-D).In the gamma band, we found a consistent and significantly lower GC in the PrL to MeAD direction in KO mice across all tasks (Fig. 5     E-H).Following this screen, we compared GC values between PrL and PVN in each direction across all sessions.We found a significant difference between C57 and KO animals in the theta GC in the PrL to PVN direction in all tasks, while in the other direction (i.e., PVN to PrL), a significant difference was only seen in the FSI task (Fig. 5I-P).Similarly, a significant difference was found in the PrL to MeAD direction for gamma GC values across all tasks, while in the other direction, such a difference was only noted in the SP task (Fig. 5Q-X).Overall, these results suggest that the PrL of KO animals exhibit hyper-synchronous neural activity with several other brain regions during social behavior.

Optogenetic stimulation of PrL pyramidal neurons abolishes emotional state preference
The data presented thus far point to the PrL in KO mice as consistently showing higher changes in theta power and synchronization with other regions during social behavior.
We, therefore, hypothesized that the hyper-active theta rhythmicity in the PrL could reflect hyper-excitability of PrL pyramidal neurons, which may elicit the behavioral impairments observed in the KO mice.To further examine this possibility, we examined the effect of stimulating PrL pyramidal neurons during the various discrimination tasks.
Accordingly, we used the AAV viral vector to transfect CamK2a-postive neurons in the PrL cortex (presumably pyramidal neurons) of both C57 and KO animals (n = 6 animals per group) with Channelrhodosin-2 (ChR2) (Fig. 6A-B).We then used an optic fiber implanted into the PrL (Fig. S2A) to apply optogenetic stimulation at either 10 or 30 Hz (or applied no stimulation at all) (Fig. S2B-C) during each of the three social discrimination tasks, which were randomized over three days of experimentation (see timeline in Fig. 6B).Interestingly, we found that in all tasks, stimulation at 10 and at 30 Hz disrupted the ability of C57 mice to prefer one stimulus animal over the other (Fig. 6C-K).Moreover, both optogenetic stimulation protocols also disrupted the ability of KO mice to discriminate between the stimulus animal and object during the SP task (Fig. 6D-E), the only social discrimination task in which non-stimulated KO mice performed normally (Fig. 6C).In contrast, optogenetic stimulation caused no behavioral changes in KO mice during the SxP and ISP tasks (Fig. 6G-K).Furthermore, both optogenetic stimulation protocols resulted in identical behavior of C57 and KO mice in all three tasks (other than a slight change in the SxP test, Fig. 6D-K).These data suggest that hyperactivity of PrL pyramidal neurons at either theta or gamma frequency causes a lack of discrimination between stimulus animals in all three tasks in a manner similar, although not identical, to the impaired discrimination exhibited by KO animals.Finally, since previous studies suggested that stimulating mPFC pyramidal neurons causes general avoidance of social stimuli and overall reduction in the tendency of a subject to engage in social interaction 28,29 , we examined which behavioral variable was changed upon optogenetic stimulation of PrL pyramidal neurons in C57 mice.We found that in all cases, investigation of the preferred stimulus animal was reduced to the level of the less-preferred one, with the latter level being mostly unaffected (Fig. 6L-Q).Since the SxP and ISP tasks involve discrimination between two stimulus animals, these results suggest that optogenetic stimulation did not cause general social avoidance, but rather the loss of behavioral preference (Fig. 6L, N, P).Such a specific effect may be caused by a reduction in the valence of the preferred stimulus, which reduces the subject's motivation to explore the stimulus.However, since the animal stimulus preferred in the SP tasks was the same type as the less-preferred stimulus animals in the other two tasks (i.e., a novel male group-housed mouse), our results contradict the possibility that a change in the absolute valence of the stimulus animal had occurred.Instead, the findings suggest that optogenetic stimulation reduced the relative valence of the preferred stimulus, and hence, the subject's motivation to explore that stimulus.We, therefore, concluded that PrL pyramidal neurons may be involved in the display of preference during a competition between two rewarding stimuli, rather than in controlling the absolute motivation to display a specific behavior.

Discussion
In the present study, we showed that Cntnap2-KO animals are impaired in their ESP behavior.Moreover, we found that even WT offspring of Cntnap2 +/-animals are impaired in this behavioral task, unlike other social discrimination tasks (such as the SP task) which they perform normally.This result, together with similar results obtained with Shank3-deficient rats (Jabarin et al., in preparation) and our previously published study of Iqsec2 A350V mice 14 , suggest that ESP behavior is especially sensitive to ASD-related mutations in mice.This may reflect the subtle behavioral, hormonal and physiological changes that distinguish an emotionally aroused animal from a relaxed animal.The same logic may explain why individuals diagnosed with ASD display specific impairments in emotion recognition, which require them to perceive and interpret subtle motion-induced changes in the facial expressions or body language of others.Thus, the ESP paradigm may be a useful tool for deciphering brain mechanisms underlying ASD-associated atypical social behavior.
Our observation that WT offspring of Cntnap2+/-animals were impaired in the ESP behavioral task is in accordance with results of other tasks found with Neuroligin3-KO mice, another well-established murine model of ASD 19,20 .Interestingly, both genes (Cntanap2 and Neuroligin3) encode synaptic proteins and both mutations seems to be associated with a modified excitatory/inhibitory (E/I) balance in the brain 18,20 .This interesting phenotype may be related to gene-environment interactions, especially as related to the effect of the gut microbiome and its spread between parents and among littermates 21,30 .
We used multisite electrophysiological recording from behaving mice 22 to explore population neuronal activity in multiple social behavior-associated brain regions during social behavior at the systems level.One intriguing observation was that KO mice exhibited generally stronger theta and gamma rhythms at baseline, even before the beginning of social interactions.This is in accordance with previously published singlecells recordings from the mPFC showing high levels of neuronal activity by Cntnap2-KO mice even before stimulus presentation 31 .Given that augmented theta and gamma rhythms are associated with internal states such as arousal and attention 32-36 , our results suggest the existence of a high level of a certain internal state in KO mice, which is in accordance with multiple studies demonstrating hyper-activity in these mice 17,18,21 .While we did not detect longer distance traveled by KO mice during the social discrimination tasks, we did observe higher levels of general investigation time, which may reflect a hyper-active or hyper-exploratory state.
Other than the fundamentally higher levels of theta and gamma rhythms, we found a higher level of social encounter-induced theta rhythmicity in KO mice, as compared to C57 mice.Moreover, we found a generally high level of theta coherence induced by social encounter in all tasks, suggesting that the various recorded brain regions are oversynchronized in KO mice during social behavior.Such globally high coherence among brain regions may cause behavioral abnormalities by masking social context-induced specific patterns of coherence between brain regions, which may be required for proper recognition of and appropriate responses to specific stimuli 37 .This result is in agreement with a recent study that used fMRI and cFos expression analyses to demonstrate macroscale functional hyper-connectivity in Cntnap2-KO mice 38 .It should, however, be noted that these authors reported reduced microscale functional connectivity among several of the brain regions recorded in the current study.This contradiction between the studies may arise from the distinct aspect of functional connectivity measured in each of them (i.e., LFP theta rhythmicity vs. BOLD fMRI signal).In any case, both studies are in accordance with other studies showing that modified synaptic connectivity in the mPFC of Cntnap2-KO mice induces modified rhythmic population activity and synchrony in this area 39 , as well as altered prefrontal functional connectivity associated with common genetic variants of the humans CANTNAP2 gene 40 .Together, these studies, which are in agreement with human studies showing a mix of hyper-and hypo-connectivity in ASD individuals 41 , support the hypothesis that the behavioral symptoms of ASD are caused by aberrant functional connectivity that may occur as a result of developmental events 42 .
We found that the PrL and PVN showed consistent and significantly augmented social behavior-induced hyper-synchrony across all paradigms.These results are very interesting, considering that the PVN is the main source of oxytocin to forebrain areas, in general, and to the mPFC, in particular 43 .Oxytocin is a neuropeptide produced solely in the hypothalamic supraoptic and paraventricular nuclei 44 and is well known for its role in regulating social behavior 45,46 .Recent studies showed that oxytocin administration can alleviate social behavior deficits exhibited by Cntnap2-KO mice 38,47 and that oxytocin is crucial for murine ESP behavior 13 .Thus, our study complements these earlier efforts by demonstrating modified synchronization between the mPFC and PVN in Cntnap2-KO mice during social behavior, thus establishing a functional link between these two regions in the context of emotion recognition in ASD.
To examine whether the behavioral deficits we observed in KO mice may be caused by hyper-excitability and hyper-synchrony of mPFC pyramidal neurons, we applied rhythmic optogenetic stimulation to excite these cells in C57BL/6J mice during social encounters.In agreement with previous studies employing a similar stimulation protocol with all PrL pyramidal neurons 29 or to specific neuronal populations innervating either the NAc 48 or the BLA 28 , we found that such stimulation impaired the SP behavior of the stimulated animals.However, unlike these previous studies, we also demonstrated that the same manipulation also abolished any preference in the SxP and ISP tasks.Notably, in all three tasks (SP, SxP and ISP), preference abolishment was induced by reducing the time dedicated by the subject to investigate the preferred stimulus animal, whereas no change was observed in the time dedicated to investigating the less-preferred stimulus, be it an object or stimulus animal.This, despite the fact that the same type of stimulus animal served as the preferred stimulus in the SP task and as the less-preferred stimulus in the other two tasks.As such, stimulation-induced reduction in investigation time was neither a general social avoidance, as suggested by previous studies 29,49 , nor stimulusspecific avoidance.Instead, it seems to be a valence-specific response 28 , expressed as a reduction in the desire to interact with a preferred stimulus.
Our observation that rhythmic optogenetic stimulation of PrL pyramidal neurons abolished the subjects' preference in all tasks, supports the idea that hyper-active and hyper-synchronous PrL neural activity, as observed by us in Cntnap2-KO mice, may underlie the impaired SxP and ISP behaviors exhibited by these animals.Our observation that these animals function normally in the SP task is most probably explained by the fact that the difference between the two stimuli (object vs. conspecific) is highest in this task, as compared to all other tasks (two conspecifics).This evident difference makes SP the least challenging of all tasks, hence causes its resilience to the naturally-induced hypersynchronous activity of PrL neurons in Cntnap2-KO mice.It should be noted that none of the ASD mouse models examined by us showed impaired SP behavior, while all of them showed impairments in ESP 14 .
Overall, our results suggest a pivotal role of fact the mPFC in social discrimination, in general, and in emotion recognition ability, in particular.They also suggest that impaired neural activity of this region, which modifies its synchronization with other social behavior-associated brain regions, such as the PVN and MeAD, is involved in the deficits exhibited by Cntnap2-KO mice in terms of emotion recognition.Such impairments in mPFC activity may also underlie similar deficits observed in other murine models of ASD 14 , suggesting a common brain pathway that integrates the effects of multiple ASDassociated mutations into a specific deficit in emotion recognition.

Box plot represents 25
to 75 percentiles of the distribution, while the bold line is the median of the distribution.Whiskers represent the smallest and largest values in the distribution.*p<0.05,**p<0.01,****p<0.0001,Paired t-test.

Figure 2 .F.
Figure 2. Cntnap2-KO male mice exhibit specific deficits in the ISP and SxP tasks

Figure 5 .
Figure 5. Theta GC values from the PrL to specific areas consistently differ between KO and C57 subjects

Figure 6 .
Figure 6.Optogenetic stimulation of PrL pyramidal neurons abolishes any preference for any given stimulus in all social discrimination tasks