Our brain processes sensory stimuli within a few milliseconds. Complex decision-making processes, on the other hand, can keep it busy for minutes. These different time scales are also reflected in the activity of the neurons. Depending on the task, these change their states rapidly or at a more leisurely pace. A study now shows that the different time scales depend on the connections of the neural networks. The results provide insights into how our brain flexibly integrates information.
The neurons in our brain fire at different rates. On the one hand, their activity is subject to intrinsic fluctuations, which are based, among other things, on different excitation decay times. Thus, the sensory areas of our brain, which must respond quickly to new stimuli, operate on a faster timescale than the association cortex and prefrontal cortex, which are involved in more complex tasks. On the other hand, the rate of fire changes for specific tasks. While these task-induced timescales are directly related to task execution, little is known about whether intrinsic timescales also change during cognitive tasks and flexibly adapt to the challenges.
Fast and slow activity
To clarify this question, a team led by Roxana Zeraati from the University of Tübingen combined data from previous publications with new computer simulations that replicate the results and allow conclusions about the underlying mechanisms. In the experiments from previous studies, macaques were trained to focus on a spot on a screen and respond to changes in various visual stimuli. Meanwhile, brainwave measurements recorded activity in the monkeys' visual cortex V4, which is responsible for this type of attention. With this type of measurement, peaks in the electroencephalogram, the so-called spikes, indicate the activity of the neurons.
"The ongoing spike activity evolved over at least two different timescales, one fast and one slow," reports the research team. In addition, fluctuations within the slow time scale could be observed during the task: if the monkey continuously focused on the given point, the slow activity of the neurons in the corresponding areas slowed down further. Should he then react to a change in the visual stimulus, this slowed down neuron activity was accompanied by the shortest reaction times.
"This may be counterintuitive, but it's actually very plausible," explains Zeraati. "A slower time scale means there is a stronger correlation between the brain's current state and its just past state. When the neurons are busy with something, they remember their own past better; and that means slowing down.”
Network structure determines the time scales
But how can a network of neurons produce these different time scales? In order to get to the bottom of the causes and mechanisms, the team created a computer simulation of the neural networks and processes and tested three different hypotheses: Are the different time scales due to the fact that the neurons involved work at different speeds? Are their different biophysical properties responsible? Or does the structure of the network determine the speed? "Our third guess turned out to be the only correct one," says Zeraati's colleague Anna Levina. "The key does not lie in the properties of individual neurons, but in the structure of the network."
Depending on how the neurons are connected to each other, different time scales are created. For example, so-called cluster networks ensure slower interconnections. "You can compare a cluster network with the European road network," explains Levina. “Any two places in Paris are very well connected; it's much more difficult to get from a village in Burgundy to a beach in Portugal.” In contrast, the airline network looks almost random: it can be difficult to get from one city to its neighboring city, but with many connecting flights you get there everywhere. "Networks that resemble airlines cannot produce such slow time scales as road network-like networks," says the researcher.
When Zeraatis and her team constructed such diverse networks on the computer, the simulations agreed well with the experimentally observed time scales. The theoretical model also explains the fluctuations during the processing of the tasks: The interactions between the neurons become slightly more efficient, so that the tempo of the neuronal events changes. Corresponding computer models could also prove helpful for future studies in order to better understand the processes in the brain. "Our experimental observations and computer models together form a basis for an investigation of the connection between network structure, functional dynamics in the brain and flexibly modulated behavior," according to the author team.
Source: Roxana Zeraati (University of Tübingen) et al., Nature Communications, doi: 10.1038/s41467-023-37613-7