Imprimir Republish


Learning to tell time

Neurons from one region of the brain train those from another to count the ticking of seconds

Inchendio / Getty Images

Even without a clock nearby, it is not difficult to count one second with reasonable precision. It’s approximately the duration of a heartbeat or a slow blink of the eye. The ability to accurately count short time intervals is often taken for granted, but it can be crucial for survival. It can mean the difference between capturing or missing prey, or crossing a street safely. But how does the brain learn to measure time? In studies with rats, researchers from the Federal University of ABC (UFABC) are helping unravel how different brain regions work together to encode the passage of these short time periods.

Their most recent findings, published in September 2022 in the journal eLife, revealed something unexpected: measuring brief intervals is not a static task performed continuously by a single area of the brain, as previous studies suggested. Instead, the team led by neuroscientist Marcelo Bussotti Reyes found that at least two regions seem to act in coordination and sequentially to perform this task.

One of these regions is the medial prefrontal cortex, a superficial layer in the front of the brain that is involved in action planning, impulse control, and rule detection. This area was found to play a role in the initial stages, learning to identify the duration of the interval. As this knowledge is firmed up, however, the timing process shifts to a deeper region: the dorsal striatum, which is responsible for the automatic execution of previously learned tasks.

At the UFABC Center for Mathematics, Computing, and Cognition, Reyes and his colleagues detected task execution shifting from one brain area to another as they analyzed neural activity in rats when learning a simple task. Within an acrylic box, the rats had to place their snout in a small porthole containing an infrared sensor, and keep it there for at least 1.5 seconds (s). If the animal withdrew its snout before that time, it would receive no reward. If it waited at least 1.5 s, it would receive a reward in the form of several licks of a sugar-water solution.

The rodents did not perform well in their initial attempts, either withdrawing their snout too early and missing the reward, or waiting too long and missing the opportunity to get more licks of the sugar solution. They only succeeded occasionally in hitting the 1.5 s mark. However, approximately one hour after the start of training, they had already learned how long they needed to wait to win the maximum reward.

This rapid improvement in task performance provided the researchers with a rare opportunity to investigate how the two brain regions known to be involved in time perception functioned at three different stages of the test: before, during, and after learning. Although time perception experiments with animals are relatively common, it is seldom possible to monitor brain activity at each discrete stage. The main obstacle is a technical one. Improvements in task performance often take days, and during that time, the electrodes implanted in the brain can shift position or lose sensitivity due to the scar tissue that forms around them. “This prevents scientists from determining if they are consistently sampling the same population of neurons,” explains Reyes. “In the current experiment, with the animals learning in less than an hour, we can be confident that we are continuously monitoring the activity of the same neurons,” he explains.

So what happens in the animal’s brain while learning the task? In the initial phase, before the animals are able to expertly count the time intervals, brain activation is observed solely in the prefrontal cortex. As the moment approaches when they can withdraw their snout from the port and receive the reward, the neurons in this area of the brain are activated more times per second. Neuroscientists refer to this pattern as a “ramping signal” due to the inclined line that appears on the graph showing the number of times each population of neurons is activated per time interval. “To our surprise, the ramping signal was present in the prefrontal cortex from the very first attempts, before the animal learned to wait the required amount of time,” says Gabriela Chiuffa Tunes, a biomedical engineer who took part in the experiments during her doctoral research under Reyes.

But as the rats gained experience, the signal shifted from the prefrontal cortex to the striatum. “At first, we attributed this to a loss in recording quality,” recalls Reyes. It was also possible that the observed effect was a mere coincidence.

But ultimately they confirmed that time encoding had indeed migrated from one region to another through a subsequent experiment. Tunes and fellow biomedical engineer Eliezyer Fermino de Oliveira administered doses of muscimol—a compound extracted from a mushroom that temporarily reduces neuronal activity—first into the prefrontal cortex and then into the striatum. When the cortex was inactivated at the beginning of the experiments, the animals were unable to learn to wait 1.5 s and performed poorly in the task. If muscimol was administered to the prefrontal cortex after the rats had become experienced, they retained their ability to obtain the reward. However, when muscimol was applied to the striatum after the rats already knew how long to wait, it reversed the learning effect, causing them to mistime the withdrawal of their snout from the porthole.

“These experiments are significant in that the authors were able to capture very early learning activity and demonstrate that, although they are directly connected to each other, the frontal cortex and the striatum play very different roles,” said Nandakumar Narayanan, a neurology researcher at the University of Iowa who was not involved in study, in an interview with Pesquisa FAPESP.

“The study insightfully suggests that, in early learning, time-related information is encoded by neurons in a highly evolved executive area of the brain—the prefrontal cortex; after learning, time is then encoded by neurons in an area traditionally associated with automatic behaviors and habit formation,” explains Adriano Tort, a neuroscientist at the Brain Institute at the Federal University of Rio Grande do Norte (UFRN). “It is an important discovery but one that needs to be corroborated by other research groups. The small number of animals observed, likely due to the complexity of the experiment, means that any statistical analysis was limited,” notes Tort.

“The current results provide evidence that time reckoning is not performed statically by a single area of the brain, at least not for short time intervals,” says Reyes. They reinforce previous findings in neuroscience research over the past few decades: that for shorter intervals, timekeeping occurs diffusely across different areas of the brain. “Our hypothesis now is that the prefrontal cortex “trains” other regions and then takes a backseat.”

1. Electrophysiological recording and manipulation of neural activity in the prefrontal cortex of rats during temporal learning (no. 16/18914-7); Grant Mechanism Direct Doctoral (PhD) Fellowship; Supervisor Marcelo Bussotti Reyes (UFABC); Beneficiary Gabriela Chiuffa Tunes; Investment R$118,514.56.
2. Characterizations of electrophysiological activity in the prefrontal-striatal pathway during temporal learning tasks (no. 16/05473-2); Grant Mechanism Master’s (Msc) Fellowship; Supervisor Marcelo Bussotti Reyes (UFABC); Beneficiary Eliezyer Fermino de Oliveira; Investment R$68,592.18.
3. Computational and systems neuroscience (no. 18/20277-0); Grant Mechanism Regular Research Grant; Principal Investigator Antonio Carlos Roque da Silva Filho (USP – Ribeirão Preto); Investment R$74,245.
4. Representation of temporal information during neural activity (no. 17/25161-8); Grant Mechanism Regular Research Grant; Principal Investigator André Mascioli Cravo (UFABC); Investment R$130,330.72.

Scientific articles
TUNES, G. C. et al. Time encoding migrates from prefrontal cortex to dorsal striatum during learning of a self-timed response duration task. eLife. Sept. 28, 2022.
BUHUSI, C. V. et al. Inactivation of the medial-prefrontal cortex impairs interval timing precision, but not timing accuracy or scalar timing in a peak-interval procedure in rats. Frontiers in Integrative Neuroscience. June 25, 2018.