Summary: A new study finds that when an action is executed the brain combines predictable and random components, processed in distinct frontal regions.
Source: Champalimaud Center For The Unknown.
Deciding when to act can be as important as deciding what to do. Even under tightly controlled laboratory conditions, the precise timing of an action often contains an unpredictable element. This apparent randomness may be beneficial, helping organisms avoid competitors and explore new possibilities.
Researchers at the Champalimaud Centre for the Unknown in Lisbon, Portugal, have shown that the timing of action execution includes both a deterministic (predictable) component and a stochastic (random) component, and that these components are represented in different frontal brain regions. Their findings were published online in the May 17 issue of Neuron.
Choosing the right moment to act is critical in many situations. Acting too early or too late can result in missed opportunities, failed outcomes, or wasted effort. Accurate timing depends on learning from past experience and adapting to current circumstances.
At first glance, adding variability or “noise” to the timing of actions might seem counterproductive. Yet unpredictability can be advantageous. For example, a skilled soccer shooter increases the chance of scoring by varying the timing and direction of a shot; if every attempt were identical and perfectly timed, defenders or opponents could easily counter it. Unpredictability also creates space for exploration and creativity, allowing organisms to discover better solutions over time.
The Champalimaud team set out to understand how the brain balances the need for well-timed actions with the benefits of unpredictability.
“We aimed to better characterize the brain mechanisms that determine when actions occur,” says Zach Mainen, who led the study. “We were particularly interested in why action timing is often highly variable—even when task conditions remain constant.”
To investigate, the researchers trained rats on a waiting task that measured patience and timing. Each trial began with an initial tone. If the rat moved immediately to the water dispenser, it received a small reward. If the rat waited until a second tone—whose timing varied randomly—it received a substantially larger reward. Over repeated trials, the animals learned to balance the learned, expected timing with a degree of trial-by-trial variability.
As anticipated, part of the animals’ waiting behavior was predictable based on experience and cues. Simultaneously, trial-to-trial variability remained prominent, revealing a meaningful stochastic component in the decision of when to move.
Separation of powers
In a second experimental phase, the team recorded neural activity from two frontal cortical areas required for the task: the medial prefrontal cortex (MPFC), which is associated with decision-making, planning and learning; and the secondary motor cortex (M2), which contributes to motor control. Both regions were necessary for choosing appropriate action timing.
Neural recordings revealed a clear dissociation. Neurons in both MPFC and M2 encoded the deterministic bias—the learned expectation of when the reward tone would arrive. However, only M2 neurons reflected the stochastic, trial-by-trial fluctuations in timing. In other words, MPFC tracked the ideal waiting time derived from experience, while M2 represented that ideal timing and also introduced variability that made each decision less predictable.
“The most surprising result was that MPFC did not mirror the random component of timing,” says Masayoshi Murakami, the study’s first author. “This functional separation between MPFC and M2 is a novel finding.”
The data also suggest a sequential interaction: deterministic signals specifying an appropriate waiting interval appear to be generated first, and stochastic variability is added downstream. According to Mainen, if variability flowed backward from M2 to MPFC, both regions would show similar stochastic signatures; instead, variability is injected later in the circuit.
These findings point to a two-stage model for action timing decisions in which circuits responsible for deterministic bias are followed by downstream circuits that inject stochastic variability. Such an arrangement allows an organism to combine reliable, experience-based timing with beneficial unpredictability.
“This interplay between optimization and variability echoes principles seen in evolution,” Mainen notes. “Our work begins to reveal how those principles operate at the level of neural circuits.”
Funding: Supported by the European Research Council, Simons Foundation, Fundação Champalimaud, Israel Science Foundation, Gatsby Charitable Foundation, Fundação para a Ciência e a Tecnologia, Uehara Memorial Foundation, and other contributors.
Source: Maria Joao Soares – Champalimaud Center For The Unknown
Image credit: Gil Costa (Champalimaud Centre for the Unknown)
Original Research: Murakami M., Shteingart H., Loewenstein Y., & Mainen Z. F., “Distinct Sources of Deterministic and Stochastic Components of Action Timing Decisions in Rodent Frontal Cortex,” Neuron. Published online May 17, 2017. doi:10.1016/j.neuron.2017.04.040
Champalimaud Center For The Unknown (2017, May 17). How the Brain ‘Plays’ With Predictability and Randomness to Choose the Right Time to Act. NeuroscienceNews. Retrieved May 17, 2017.
Abstract
Distinct Sources of Deterministic and Stochastic Components of Action Timing Decisions in Rodent Frontal Cortex
Highlights
• Medial prefrontal cortex and secondary motor cortex are required for a waiting task.
• Medial prefrontal cortex encodes deterministic bias in action timing.
• Secondary motor cortex encodes stochastic variability in action timing.
• Distinct timescales of neural dynamics in frontal cortex reflect different functions.
Summary
Action selection and timing are influenced by deterministic factors, such as sensory cues and internal state, as well as by effective stochastic variability. Although many theoretical models incorporate stochastic choice mechanisms, their neural origins have been unclear. This study examined how frontal cortical circuits determine action timing in a rat waiting task. Electrophysiological recordings from medial prefrontal cortex (mPFC) and secondary motor cortex (M2)—both necessary for the behavior—revealed a functional dissociation: both areas encoded deterministic biases in timing, but only M2 neurons displayed trial-by-trial stochastic fluctuations. These differences corresponded to distinct neural dynamics timescales in the two regions and support a two-stage model in which downstream circuits inject stochastic components after deterministic bias signals are formed.
“Distinct Sources of Deterministic and Stochastic Components of Action Timing Decisions in Rodent Frontal Cortex” by Masayoshi Murakami, Hanan Shteingart, Yonatan Loewenstein, and Zachary F. Mainen. Neuron. Published online May 17, 2017. doi:10.1016/j.neuron.2017.04.040