scholarly journals Sequential sampling from memory underlies action selection during abstract decision making

2021 ◽  
Author(s):  
S Shushruth ◽  
Michael N Shadlen

The process of deciding what a sensory stimulus is and how to act on that decision seem distinct, yet they appear to be coupled at the neural level. Neurons in the parietal cortex of monkeys represent both the integration of evidence toward a decision and the behavior used to report the decision. This raises the possibility that monkeys evaluate sensory percepts in terms of their motor affordances rather than their abstract identity. It is not clear how monkeys can evaluate sensory percepts when unaware of the motor actions they bear upon. We investigated this by training monkeys to make perceptual decisions about the direction of motion in a noisy random-dot display. They learned to associate leftward and rightward with two colors, and to select from a pair of colored targets, which were displayed after the motion at unpredictable locations. Surprisingly we found that monkeys postpone decision formation until the pertinent motor actions are revealed. Neurons in parietal cortex represent the accumulation of evidence sampled from short term memory of the motion display. The findings demonstrate that abstract decisions are framed in terms of their motor affordances and highlight the capacity for integration of evidence from memory.

2020 ◽  
Vol 10 (6) ◽  
pp. 2000 ◽  
Author(s):  
Jihwan Park ◽  
Mi Jung Rho ◽  
Hyong Woo Moon ◽  
Ji Youl Lee

It is particularly desirable to predict castration-resistant prostate cancer (CRPC) in prostate cancer (PCa) patients, and this study aims to predict patients’ likely outcomes to support physicians’ decision-making. Serial data is collected from 1592 PCa patients, and a phased long short-term memory (phased-LSTM) model with a special module called a “time-gate” is used to process the irregularly sampled data sets. A synthetic minority oversampling technique is used to overcome the data imbalance between two patient groups: those with and without CRPC treatment. The phased-LSTM model is able to predict the CRPC outcome with an accuracy of 88.6% (precision-recall: 91.6%) using 120 days of data or 94.8% (precision-recall: 96.9%) using 360 days of data. The validation loss converged slowly with 120 days of data and quickly with 360 days of data. In both cases, the prediction model takes four epochs to build. The overall CPRC outcome prediction model using irregularly sampled serial medical data is accurate and can be used to support physicians’ decision-making, which saves time compared to cumbersome serial data reviews. This study can be extended to realize clinically meaningful prediction models.


Sign in / Sign up

Export Citation Format

Share Document