Making the Best Choice Between Multiple Options

Post by Deborah Joye

What's the science?

The ability to make decisions is an important part of survival. But how do individual neurons or the brain as a whole calculate what the best choice is? Researchers have described how choosing between two things might occur, but whether that process is the same for choosing between more than two options is not known. One possible decision-making model involves a “race” concept where evidence in favor of each option is accumulated over time until a decision threshold is reached, triggering a choice. But this model assumes that each option accumulates evidence independently and that the decision criteria remains the same across time and options. The nervous system is also much more complex than the “race” model suggests. For example, possible options can influence the perceived value of other choices, a process called value normalization. The nervous system may also involve a “global urgency signal” which decreases the amount of evidence needed to trigger a decision as time elapses. This week in Nature Neuroscience, Tajima and colleagues use tools from dynamic programming to present a model for optimal decision-making between multiple choices. The authors describe an extended race model that considers how alternative choices interact over time and how a global urgency signal increases the likelihood of a decision as time elapses.

How did they do it?

The authors built a mathematical model that describes how a decision-maker accumulates evidence for each choice option over time, how choice options interact with one another, and when the decision-maker should stop accumulating evidence and make a choice. The evidence accumulation part of the model involves summing together noisy moment-to-moment evidence for each choice over time. To identify the optimal strategy to stop accumulating evidence, the authors designed a 'value function’ which is a mathematical function that calculates the expected reward for being in a certain state at a given time. Their value function considers maximum expected reward minus the cost of accumulating evidence over time (for example, a metabolic cost associated with continuing to gather evidence) and compares the value of deciding immediately versus accumulating more evidence and deciding later. Solutions to these equations and the points at which they intersect to produce a multi-dimensional space with decision boundaries which, once crossed, indicate that option is chosen. Importantly, the authors’ value function considers the passage of time and models how decision boundaries change as time elapses. Overall, this leads to a fairly complex optimal decision-making strategy. To propose a simpler, close-to-optimal strategy, the authors use the insights gained from the dynamic programming solution to design a model neural circuit to implement their optimal decision-making policy and describe how their modelled data matches collected behavioral data.

What did they find?

When the authors simulated the basic “race” model versus “race” models that considered how options affect one another over time (normalization), how the decision boundaries change over time (urgency signal) or both, they found that the best model requires both normalization and urgency. Interestingly, they found that the inclusion of normalization alone improved reward rate more than the inclusion of urgency alone, demonstrating the importance of considering how the value of different choice options interact over time. The authors also found that data output from their model closely resembled previous physiological and behavioral findings. For example, when their model was optimized for maximum reward rate, the authors found that average activity across all accumulating neurons increased over time, replicating the physiological urgency signal that increases the likelihood of a decision over time. The authors also found that increasing the number of options in their model resulted in a decrease in average neural activity and increased reaction time. This replicates physiological and behavioral data demonstrating that more options mean the decision-maker (whether it’s a person or a neuron) accumulates evidence more slowly and takes more time to decide. Their optimal reward model also replicates seemingly “irrational” behaviors in humans and animals such as estimating the value of two top choices depending on the value of a third option, even if the third option is not valued enough to ever be chosen (violating the idea of independence of irrelevant alternatives) or that adding extra options can increase the probability of selecting an existing option (violating the regularity principle).

Deb (3).png

What's the impact?

This study provides a cohesive model for optimal decisions between more than two options. Importantly, the model replicates data from previous behavioral studies as well as seemingly “irrational” choice behaviors that other decision models do not explain. These findings provide a new perspective on decision-making that considers changes over time and brings together contrasting models. This model also produces interesting predictions that can be tested behaviorally and physiologically in future experiments.

optimal_decisions_quote.jpg

Tajima et al., Optimal policy for multi-alternative decisions, Nature Neuroscience (2019). Access the original scientific publication here.

Spatially Specific White Matter Tracts are Associated with Reading and Math

Post by Shireen Parimoo

What's the science?

Reading and math involve similar cognitive processes like working memory and verbalization, and math and reading-related disabilities tend to co-occur. Previous research examining reading has identified white matter tracts in the brain that are related to performance on reading tasks, such as the arcuate fasciculus and the inferior longitudinal fasciculus. However, less is known about the white matter tracts associated with math. It is also unclear whether there is an overlap in the structural properties of white matter tracts associated with reading and math ability. This week in Nature Communications, Grotheer and colleagues used multimodal magnetic resonance imaging (MRI) techniques to identify the shared and distinct structural correlates of the cognitive processes involved in reading and addition.

How did they do it?

Twenty adults completed a reading, adding, and a control color task while undergoing functional MRI scanning. The task stimuli were morphs that could be perceived as a number or a letter (e.g. the stimulus for the letter “S” could be perceived as the number “5”), which allowed the visual input to remain constant across the different tasks. A series of four stimuli were presented consecutively. In the reading task, participants had to indicate the word spelled out by the letters; in the addition task, participants had to add up the numbers; in the color task, participants had to indicate the color of the stimuli at the end of the stimulus sequence. The fMRI data was used to determine which brain regions were activated during the reading task, during the addition task, or during both tasks. These brain regions of interest were identified for each participant and co-located white matter tracts were then analyzed.

Diffusion MRI and quantitative MRI scans were performed to investigate the connectivity and microstructure of white matter, respectively. First, the authors used constrained spherical deconvolution (a method used to model the orientation of white matter fibers) on the diffusion MRI data to create a structural connectome, or a map of the white matter pathways connecting brain regions. They then applied an automated algorithm to identify the major white matter pathways (also called fascicles) in the brain, including the arcuate fasciculus (AF), the posterior arcuate fasciculus (pAF) and the superior longitudinal fasciculus (SLF). The white matter fascicles were intersected with the functionally-defined ROIs to localize the tracts associated with reading and math. This allowed the researchers to examine which white matter tracts support the connectivity within and between the reading and math networks. Finally, quantitative MRI was used to estimate the myelination of the white matter tracts connecting regions of the reading and math networks. In general, greater myelination is associated with more efficient neuronal transmission in the brain.  

What did they find?

Reading and math activated largely separate but neighbouring brain regions. For instance, the occipitotemporal sulcus, the superior temporal sulcus, and the inferior frontal gyrus were active during the reading task, whereas the addition task activated the inferior temporal gyrus, the inferior post-central sulcus, and the intraparietal sulcus. Both tasks also activated distinct subregions within the supramarginal gyrus. Across the brain, reading- and math-specific regions were connected to their respective network by the SLF, the AF, and the posterior AF. The SLF and AF connected the prefrontal regions of the math and reading networks, such as the inferior frontal gyrus (reading) and the post-central sulcus (addition), to temporal and parietal regions of each network. Further, the posterior AF connected the temporal regions active during the two tasks, such as the occipitotemporal sulcus (reading) and the inferior temporal gyrus (adding) to the parietal regions of each network. Thus, the same white matter fascicles support the math and reading networks, even though the brain regions themselves are largely distinct. Crucially, though, analogous to distinct lanes in a highway, math and reading-related white matter tracts were found to run in parallel, segregated sub-bundles within the shared fascicles. The specific sub-bundles, or branches, of the SLF and AF involved in reading were located more inferiorly in the brain than those involved in addition. Moreover, the branches of the SLF and AF involved in reading were more heavily myelinated than those associated with the addition network, suggesting greater efficiency of neuronal transmission within the reading network.

math_reading_img.jpg

What's the impact?

This study is the first to establish that spatially distinct branches of the same white matter fascicles are associated with the reading and math networks in the brain. These findings suggest that the ability to read and perform mathematical operations might develop independently, despite shared cognitive processes and a high rate of comorbidity of their associated learning disorders. This has important implications for future research exploring the relationship between white matter properties and math- and reading-related abilities.  

math_reading_quote.jpg

Grotheer et al. Separate lanes for adding and reading in the white matter highways of the human brain. Nature Communications (2019). Access the original scientific publication here.

Deep Brain Stimulation Normalizes Brain Activity in Parkinson’s Disease

Post by Elisa Guma

What's the science?

Deep brain stimulation (DBS) is an effective and established treatment for Parkinson’s disease, wherein electrodes are implanted in a targeted brain area in order to relieve certain symptoms such as tremor, stiffness and rigidity, and impaired gait. The mechanism by which DBS is thought to improve symptoms is still not fully understood. It was previously thought that improvements were solely due to localized stimulation of specific brain regions, however, they may be due to more global changes in functional brain networks. This week in Brain, Horn and colleagues investigated the effects of DBS on functional brain networks of patients suffering from Parkinson’s disease.

How did they do it?

The authors acquired resting-state functional magnetic resonance imaging (rs-fMRI) data in 20 Parkinson’s patients who underwent surgery to place DBS electrodes in the subthalamic nucleus (STN), and 14 healthy, age-matched controls. RS-fMRI is a technique that measures fluctuations in blood oxygenated level-dependent (BOLD) signals in the brain, which allows for a measure of intrinsic associations between the brain activity of specific regions based on the correlation of signal over time between those brain regions. Patients were first measured with their electrodes turned on, and after a short break, were scanned again with their electrodes turned off in order to see how the electrical stimulation affected global brain activity.

The data was processed using state of the art software, Lead-DBS, which allowed for the localization of the DBS electrodes, as well as analysis of brain volume and activation, with careful regard for artefacts due to motion during scans and metal from the electrodes. This allowed the authors to analyze how the electric field of the DBS-electrodes modulated brain activity in a key motor network referred to as the basal ganglia-cerebellar-cortical loops. These loops include the sensorimotor functional zones of the cortex, striatum, thalamus, internal and external globus pallidus, substantia nigra, and cerebellum. They compared functional brain networks in the DBS-on and -off conditions to those of healthy controls. Further, they investigated how electrode placement modulated the patterns in brain activity they observed.

What did they find?

First, the authors found that the accuracy of the electrode placement within the STN determined the strength of connectivity between the STN and the supplementary motor area (a motor network region); the better the placement, the stronger the connectivity between these two regions. Next, they found that connectivity maps between the volume of tissue activated around the STN and the motor network were most similar between DBS-on conditions and healthy controls, suggesting that DBS electrode activity might normalize brain networks towards healthy controls. This was also affected by the electrode placement. Finally, the authors found that connectivity in the DBS-on group was increased in the motor network (between the thalamus and cortex), with a decrease in basal ganglia connectivity (striatum to cerebellum, STN, and STN to globus pallidus).

DBS.png

What's the impact?

This study is one of the first to demonstrate the feasibility of conducting rs-fMRI in DBS implanted patients. They identify that DBS has a significant effect on brain connectivity throughout the motor network and that these changes were strongly dependent on correct electrode placement. The findings are promising evidence for the use of invasive neuromodulation. Further, DBS provides a framework within which to study how brain networks change in response to targeted stimulation, which could be applied to other populations undergoing DBS treatment, such as those with depression, obsessive-compulsive disorder, or eating disorders.

DBS_quote.jpg

Andreas Horn et al. Deep brain stimulation induced normalization of the human functional connectome in Parkinson’s disease. Brain (2019). Access the original scientific publication here.