On September 29, 2025, Simons Foundation Executive Vice President for Autism and Neuroscience Kelsey Martin welcomed over 50 attendees to the annual meeting of the Autism Rat Consortium (ARC) in New York City. Since 2021, the ARC has studied neural circuits and behavior in multiple rat models of autism spectrum disorder (ASD) generated by the Simons Foundation Autism Research Initiative (SFARI) and the Simons Initiative for the Developing Brain (SIDB) in Edinburgh.

SFARI Senior Scientific Officer Brigitta Gundersen made the case for studying rats relative to mice: Their larger size makes it is easier to study brain development; they can be trained to perform complex tasks, which helps probe cognition; their rich social repertoire provides an opportunity to measure social behaviors that may be more informative in understanding ASD’s social deficits; and their outbred genetic background could help make sense of phenotypic variability that results from the same genetic mutations. Members of ARC conduct research on a wide range of questions in ASD rat models, including social behavior and its associated neurophysiology, sensory processing, navigation, and learning and memory.

Peter Kind of the University of Edinburgh and SIDB noted that SIDB has made several rat models, including Syngap1, Cdkl5 and Grin2a. SIDB has also developed a behavioral phenotype pipeline, which is used to characterize each rat line. SIDB has also developed conditional rat lines that allow researchers to discriminate between the actions of a gene during neurodevelopment in young animals versus neuromaintenance in adults. For example, turning the fragile X-related gene Fmr1 off in juveniles, then turning it back on in adulthood can help isolate any long-lasting effects of Fmr1 loss during brain development.

Kind elaborated on the behavioral phenotyping pipeline, which includes neonatal reflexes, tests of object location memory and object recognition, water maze navigation, active place avoidance, auditory fear conditioning, prey capture, sensory processing tasks, and play behavior in juveniles band adult social interaction. The behavioral data for each rat line is summarized in a data sheet available to the public.

While the pipeline relies on task-based readouts, SIDB has also begun to characterize natural rat behavior in a modular ‘Habitat’ environment that mimics the warrens in which rats live in the wild. Consisting of 16 interconnected modules, the Habitat can house 20 to 80 animals, which freely move from one module to the next. Their movements and interactions can be tracked, through scoring videos or automatically with radio-frequency identification (RFID) chipping. Preliminary analyses show circadian rhythm-related movement patterns and differences in habitat space use among rats. The team is developing measures to quantify aspects of social interaction such as how often two particular rats are co-localized.

Motivated by differences in sensory perception and navigation found in ASD, Paul Dudchenko of the University of Stirling presented his work on the head direction system in rats. Found in mammillary bodies, thalamus and cortex, head direction neurons are tuned to the head pointed in a particular direction, which could make them instrumental in building internal representations of the outside world. Their responses are anchored to salient visual landmarks, but can also be shaped by self-motion via vestibular input. Using LED screens to rotate visual cues, Dudchenko has found that Fmr1 adult rats rely on visual cues alone and that their self-motion coding is reduced. This is consistent with their normal learning of landmarks and their impaired self-motion navigation. Ongoing experiments include tracking head-direction tuning during development and in learning tasks that probe the malleability of landmark visual cues. Future work will test the sensitivity of head direction cells to visual or vestibular cues in other rat lines, including Scn2a, Arid1b and Grin2b. During questions, Dudchenko noted that the vestibular system in Fmr1 rats seems normal with respect to gross motor skills, like walking.

The next two talks explored behavioral and neural responses of rats lacking a working copy of the sodium channel-encoding Scn2a gene. In his talk, David Kastner of the University of California, San Francisco (UCSF) noted that while Scn2a haploinsufficiency alters neural excitability, function and plasticity¹, it remains unclear how this leads to the complex ASD phenotype. Kastner suggested that part of the answer may lie in differential use of brain regions; this could bias an animal toward different behavioral phenotypes and explain individual variability. Based on hippocampal lesions in control rats, Kastner used a data-driven behavioral analysis method² to classify sequences of choices on a six-arm track learning task as hippocampal-dependent or hippocampal-independent. Learning of Scn2a-haploinsufficient rats showed a shift towards the use of more hippocampal-independent sequences than controls rats, suggesting reduced reliance on the hippocampus for their behavior.

Kevin Bender of UCSF noted that the loss of Scn2a may be affecting homeostatic mechanisms in cortical neurons, as there are signs of hyper-excitability in local field potential (LFP) and electroencephalography (EEG) recordings. This may promote the absence-like seizures and fractured sleep cycles found in Scn2a rats. Bender also presented work from collaborator Loren Frank (UCSF) showing that in navigation tasks that rely on experience-guided decisions, Scn2a rats seem more rigid in their choices than control rats, and at choice points in a maze, they fail to show hippocampal activity that encodes actual and alternative (potential future) locations³. There may be a link between hyper-excitability and navigation difficulties: Rats with more seizures were less likely to show activity representing alternative locations. To try to rescue some of these deficits, Bender and Frank are working with Nadav Ahituv of UCSF to adapt CRISPR-activation techniques to rats to increase expression of the intact Scn2a allele⁴.

Shantanu Jadhav of Brandeis University described his methods for probing cooperative behavior among rats, which could highlight deficits in ASD rat models⁵. The task involves two rats, each in its own three-armed track; these tracks face each other, with a transparent barrier between them. The rats learn to go to the ends of the complementary arms and nose-poke simultaneously to gain a reward, even when they are only rewarded 50 percent of the time. Fmr1 rats can learn this task, but compared to control rats, they engage less in the task, show slower learning curves, and don’t achieve the same level of performance. Other control experiments suggest this stems from social learning deficits, rather than sensory processing: Fmr1 rats follow a reactive “follow the leader” strategy, whereas control rats develop a prediction of what their partner will do, and update the internal model as necessary. In discussion, Jadhav noted that he thought the partner rat was a salient stimulus to Fmr1 rats, though kinship may matter, as they will often ignore non-kin rats.

In her talk, Gina Turrigiano of Brandeis University explored different mechanisms of learning in rat models of ASD, using a cricket-hunting behavior that rats naturally display but in which they improve significantly with training. After juvenile rats learn this skill, Turrigiano finds a slow increase in baseline firing rate among primary visual cortex (V1) neurons, which could reflect the establishment of a new baseline level of network activity⁶. Learning is also accompanied by an increase in spine density on excitatory neurons in layers 2/3 and 5, which relies on excitatory synaptic homeostatic plasticity; blocking this plasticity mechanism degraded hunting skill in rats. Though Shank3 and Scn2a rat models could learn the task, Shank3 rats were slower and struggled with retention. In vitro, there were signs of impaired homeostatic plasticity in Shank3 neurons, such that reducing their synaptic input did not cause an increase in intrinsic excitability. In contrast, neurons from Scn2a rats seemed to be preadapted to an intrinsic increase in excitability; these physiological mechanisms could contribute to the learning differences.

Davide Zoccolan of the Scuola Internazionale Superiore di Studi Avanzati described his characterization of high-level visual perception in Scn2a rats. Similar to what has been reported in ASD, Scn2a rats show increased sensitivity to high spatial frequency stimuli compared to controls. Specifically, a high frequency priming stimulus could more strongly bias orientation perception in Scn2a rats than in wild type rats. In a task that probed perception of larger, global structures at the expense of local features, Scn2a rats did worse than wild type rats, even when primed with a stimulus to bias them toward global feature detection. This result is consistent with local feature dominance reported in ASD. Preliminary data indicate that Scn2a rats have an intact optomotor response, which relies on subcortical regions; thus, their high spatial frequency sensitivity may arise in cortical processing. Zoccolan is exploring this with recordings in higher visual regions, including extrastriate areas involved in shape processing, and with immediate early gene activation to mark neural activity.

Marino Pagan of the University of Edinburgh and SIDB presented his experimental paradigm to study decision-making and cognition in rats, which he is now applying to Fmr1 rats. In one task, rats learn to switch their attention between two different features of an auditory stimulus, then form decisions about the relevant feature. Control rats can learn this with automated training, but Fmr1 rats are impaired. Neural recordings in prefrontal cortex show that a heterogeneous population of neurons encodes the upcoming choice along a fixed ‘choice axis’, reflecting the accumulation of incoming information as the animal makes a decision⁷. For Fmr1 rats, preliminary data shows that their neural activity (and decision-making) weighs the initial stimuli more strongly than later ones.

Pagan is taking a similar approach to study a social interaction task involving two rats trying to push each other out of a tube. Neural recordings in prefrontal cortex show activity with population dynamics along a single axis that reflect whether the animal is being pushed or is pushing.

Ann Kennedy of Scripps Research described work performed by collaborator Bence Ölveczky’s lab⁸ studying social interactions in rats as a means toward understanding the cognitive processes and neural circuit phenotypes associated with social behaviors. Using phenotypic characterization of rat behavior, they can detect and track stereotyped actions of individual animals, as well as joint behavior between interacting animals, including touch. Kennedy is working on computational models that can predict an animal’s future actions based on their environment. The resulting “sensorimotor policies” would define how an animal takes in information, then produces social interactions accordingly. These kinds of distilled guidelines could help understand the sensory cues and environment that shape a rat’s actions, identify associated neural circuits important to social behavior, and highlight circuits likely to be altered in ASD rat lines.

In his talk, Kennedy and Ölveczky’s collaborator Naoshige Uchida of Harvard University focused on dopaminergic systems in the brain that are involved in flagging social valence. Specifically, threat and avoidance are mediated by the tail of the striatum (TS), which receives dopamine inputs from the substantia nigra pars lateralis, whereas reward and approach are mediated by the nucleus accumbens (NAc), which receives dopamine inputs from the ventral tegmental area (VTA). Uchida hypothesized that sociability is regulated by a balance between NAc and TS dopamine signaling⁹. To characterize this balance in individual animals, including ASD rat models, his lab is recording dopamine signals in NAc and TS during social interactions, which are quantified with automated gesture-tracking¹⁰.

The last series of talks highlighted the research plans of ARC investigators newly funded through the ARC 2.0 RFA. From the University of Maryland School of Medicine, Margaret McCarthy and Steffen Wolf made the case for studying play behavior in rats as a way to get closer to the neurobiology underlying social deficits in ASD. Juvenile rats are highly motivated to play, and during pair-wise “play dates” they exhibit a set of play elements, such as pouncing, pinning, and boxing. The neural circuitry mediating play involves the medial prefrontal cortex and medial amygdala (MeA), which belong to a social behavior network in adults. McCarthy’s lab will explore how ASD-related mutations affect rat play; earlier results have found reduced play in Nrxn1 rats¹¹. The lab will use automated gesture-tracking to capture and quantify play¹². Preliminary analyses show this method can pull out recognizable play elements, and distinguish differences in male and female play patterns. Wolff will record neural activity related to play, through chronic recordings of MeA. Though technically challenging, chronic recordings have been obtained from playing rats without movement artifacts.

In his talk, Tim Hanks of the University of California, Davis focused on dopamine’s role in ASD-related behaviors. Responsible for reward, movement, action selection, motivation, and behavioral flexibility, dopamine has also been found to be dysregulated in ASD. Hanks has developed a task that can distinguish value-based decision-making over rigid, perseverative choices, while simultaneously measuring dopamine release in the striatum or the prefrontal cortex (PFC) with fiber photometry. Hanks will explore the relationship between dopamine release in the striatum and value-based or perseverative behavior in wild type and Fmr1 rats. He hypothesizes that striatal dopamine release will be associated with perseverative behavior in Fmr1 rats relative to controls. He will also apply this framework to experiments in PFC, in which dopamine release is more sustained than in the striatum.

João Couto and collaborator Anne Churchland of the University of California, Los Angeles plan to study the brain-wide changes that accompany perceptual decision-making in multiple ASD rat models. This will involve a multisensory behavioral task similar to the one used by Pagan, in which rats are presented with a stream of auditory, visual or multisensory stimuli. Rats learn to reliably report whether the rate of this stream is higher or lower than a memorized category boundary. ASD rat models are hypothesized to have difficulties with this task, given documented changes in sensory processing. While the rats are engaged in this task, Couto and Churchland will also employ brain-wide electrical recordings using Neuropixels, a device that enables large scale, simultaneous, chronic neural recordings¹³. Sampled regions will include sensory and association areas, as well as higher cognitive areas of the cortex, in order to get a sense of brain-wide activity changes. They will attempt to rescue any detected circuit deficits using optogenetic activation or inhibition, which could demonstrate a causal role for a brain region in the task, as well as offer up potential therapeutic strategies.

James Dooley of Purdue University outlined his plans to study the sleep-dependence of motor circuit development in neonatal rats, which may be disturbed in ASD rat models. Early in life, movement is driven by the red nucleus in the brain stem, then later is replaced by motor cortex. This switch requires REM sleep, as well as red nucleus-driven motor twitches, which initially evoke sensory responses in the primary motor cortex (M1)¹⁴. Dooley will characterize this transition in Fmr1 rats, using multi-electrode probes to record from M1 and the red nucleus throughout development. Dooley hypothesizes that during REM sleep, motor cortex projection neurons can selectively inhibit the red nucleus, through synapses onto parvalbumin-positive (PV+) interneurons; this suppression would then give M1 an opportunity to produce its own movements through the corticospinal tract. In Fmr1 rats, PV+ interneurons may show decreased activity during REM sleep, which would allow the rubrospinal pathway to continue to dominate movement production longer than usual.

In the final talk, Ben Auerbach of the University of Illinois Urbana-Champaign outlined his plans to examine auditory processing differences in multiple ASD rat models, given prevalent disruptions in sensory processing seen in ASD. Using Fmr1, Tsc2 and Scn2a rats, he will characterize auditory processing behaviorally and with electrophysiology to probe detection and sensitivity thresholds, frequency discrimination, and temporal processing in multiple brain regions. He will also monitor auditory system development longitudinally with assays such as acoustic startle reflex, innate sound avoidance, and auditory brainstem recordings. An operant sound detection task has already illuminated differences in Fmr1 rats: Although they have similar hearing thresholds as control rats, Fmr1 rats show increased sensitivity to loudness¹⁵, reduced sensitivity to stimulus duration, and difficulties in discriminating tones of similar frequency. Recordings at different points of the auditory pathway finds evidence for central auditory gain enhancement in Fmr1 rats, which may increase sound sensitivity at the expense of fine-feature discrimination.

In closing, Gundersen thanked attendees for participating in the meeting and encouraged them to continue their conversations, either virtually or in person.

1. Spratt P.W.E. et al. Neuron 103, 673–685 (2019) PubMed
2. Kastner D.B. et al. bioRxiv (2024) Preprint
3. Comrie A.E. et al. bioRxiv (2024) Preprint
4. Tamura S. et al. Nature 646, 983–981 (2025) PubMed
5. Shukla A. et al. bioRxiv (2025) Preprint
6. Leman D.P. et al. bioRxiv (2025) Preprint
7. Pagan M. et al. Nature 639, 421–429 (2025) PubMed
8. Klibaite U. et al. Cell 188, 2249–2266 (2025) PubMed
9. Tsutsui-Kimura I. et al. Nat. Neurosci.</em 28, 795–810 (2025) PubMed
10. Klibaite U. et al. Cell 188, 2249–2266 (2025) PubMed
11. Kight K.E. et al. Behav. Neurosci. 135, 782–803 (2021) PubMed
12. Klibaite U. et al. Cell 188, 2249–2266 (2025) PubMed
13. Melin M.D. et al. bioRxiv (2024) Preprint
14. Dooley J.C. and Blumberg M.S. eLife 7, e41841 (2018) PubMed
15. Auerbach B.D. et al. Neurobiol. Dis. 161, 105541 (2021) PubMed