Summary: The medial prefrontal cortex persistently encodes value based decision variables.
Source: Johns Hopkins Medicine
Mammal brains — including those of humans — store and recall impressive amounts of information based on our good and bad decisions and interactions in an ever-changing world. Now, in a series of new experiments with mice, scientists at Johns Hopkins Medicine report they have added to evidence that such “decision-based” memories are stored in very particular parts of the brain.
“Figuring out where and how long animal brains store information about past choices can help us more broadly understand models of decision-making,” says Jeremiah Cohen, Ph.D., assistant professor of neuroscience at the Johns Hopkins University School of Medicine. Eventually, he says, such models could lead to better treatments for addiction and other disorders marked by flawed or disordered decision-making.
A description of the studies was published July 4 in the journal Neuron.
For the experiments, Cohen and his colleagues at Johns Hopkins used various odors to train 36 mice to choose between two tubes that deliver water when licked and which were randomly controlled by a computer to turn the “faucets” on or off. When one tube was turned on, the other was turned off, and then, they’d switch. The experiment was designed to mimic how mice forage for water in the wild and learn where to return for the liquid.
The mice didn’t know, with certainty, whether they would receive water when they made a particular choice, says Cohen. To find the water reward, they had to remember their recent choices and outcomes, he explains.
As expected, Cohen says, the mice more often explored and licked from water tubes that were turned on more frequently, maximizing the amount of water they received. Occasionally, they also explored water tubes that were turned off, licking them once in a while. When a tube with plentiful water was turned off, the mice switched to the other tube that was newly turned on, says Cohen. The scientists also found that the mice waited to make their water tube choices for up to half a minute.
Cohen says previous studies over the past decade have suggested that decision-based memories are stored in the prefrontal cortex — the seat of complex behavior, personality, and learning.
To figure out where in their brains the mice stored the water memories, a member of Cohen’s team, Bilal Bari, attached microelectrodes to the brains to measure the firing rate of so-called “action potentials” crossing the membranes of more than 3,000 neurons in the prefrontal cortex.
According to Cohen, the firing rate of the neurons’ action potentials in a mouse represents the internal cognitive estimate of the choice between the two water tube options — basically, the mouse thinking about which tube to lick.
A baseline action potential rate in prefrontal cortex neurons is about two to 10 spikes per second. But when the mice were believed to be thinking about one tube choice versus the other, the scientists found action potential rates of an average of about 20 spikes per second in nearly 80% (2,401 of 3,073) of prefrontal cortex neurons.
They also found that the longer the mice took to decide between the two tubes, the action potential rate in prefrontal cortex neurons started to slow down. After 15 seconds without licking a tube, action potential rates were down to half their peak rate. “It’s possible that the mouse is forgetting how valuable the particular environment is,” says Cohen.
“This data gives us a window into how neurons are storing these decision-based memories,” says Cohen.
“Not only can we see the changes in action potential firing rates, but we can quantify them too.”
The researchers also measured wherein the mouse brains the prefrontal cortex neurons extended to other parts of the brain, signaling structures elsewhere. They found increased electrical signaling to the dorsomedial striatum, a structure inside the basal ganglia, a region in the middle of the brain linked to movement.
Cohen says his team plans further study on where and how neurons connect from the prefrontal cortex to other parts of the brain that control movement. This information would likely shed light on how fast animals including humans learn from their experiences in the world. “We need to learn quickly in a volatile world, and figuring out more of how brains do that will offer insight into how such learning can be adjusted,” Cohen says.
Funding: Funding for the study was provided by the National Institutes of Health’s National Institutes of Mental Health (F30MH110084), the National Institute on Drug Abuse (R01DA042038), the National Institute of Neurological Disorders and Stroke (R01NS104834, P30NS050274), a Klingenstein-Simons Fellowship Award and the National Alliance for Research on Schizophrenia and Depression.
In addition to Bari and Cohen, the following scientists from Johns Hopkins contributed to the research Cooper Grossman, Emily Lubin, Adithya Rajagopalan and Jianna Cressy.
Johns Hopkins Medicine
Rachel Butch – Johns Hopkins Medicine
The image is adapted from the Johns Hopkins Medicine news release.
Original Research: Closed access
“Stable Representations of Decision Variables for Flexible Behavior”.
Bilal A. Bari, Cooper D. Grossman, Emily E. Lubin, Adithya E. Rajagopalan, Jianna I. Cressy, Jeremiah Y. Cohen.
Journal of Psychiatric Research. doi:10.1016/j.neuron.2019.06.001
Stable Representations of Decision Variables for Flexible Behavior
• Medial prefrontal cortex (mPFC) persistently encodes value-based decision variables
• Relative value signals are stable, and total value signals decay
• Persistent decision variables are weakly represented in premotor cortex
• mPFC projections to dorsomedial striatum persistently represent decision variables
Decisions occur in dynamic environments. In the framework of reinforcement learning, the probability of performing an action is influenced by decision variables. Discrepancies between predicted and obtained rewards (reward prediction errors) update these variables, but they are otherwise stable between decisions. Although reward prediction errors have been mapped to midbrain dopamine neurons, it is unclear how the brain represents decision variables themselves. We trained mice on a dynamic foraging task in which they chose between alternatives that delivered reward with changing probabilities. Neurons in the medial prefrontal cortex, including projections to the dorsomedial striatum, maintained persistent firing rate changes over long timescales. These changes stably represented relative action values (to bias choices) and total action values (to bias response times) with slow decay. In contrast, decision variables were weakly represented in the anterolateral motor cortex, a region necessary for generating choices. Thus, we define a stable neural mechanism to drive flexible behavior.