Decision making and sequential sampling from memory

(Shadlen & Shohamy, 2016)

In this article, the authors consider the trade-off between speed and accuracy when making decisions. They draw on examples of sequential sampling and accumulator models to explain simple sequential decisions but ask how value-based decisions accumulate evidence. They argue that memory and past experiences inform prospective reasoning and decisions.

The authors discuss this in three sections:

1. Reviewing of existing data regarding accuracy and timing of perceptual and value decisions.
2. Reviewing of existing information on the role of memory on value decisions
3. Presentation of a “working framework for neurobiological mechanisms supporting circuit-level interactions by which sampled evidence from memory can influence value-based decisions and actions.”

 

1. Accuracy and timing of perceptual data is discussed with reference to the random dot motion task and the bounded evidence accumulation framework (drift diffusion. Accuracy and timing of value-based decisions are described as decisions where “subjective preferences and predictions about the subjective value of each option…rather than on objective data bearing on a proposition about a sensory feature” (Shadlen & Shohamy, 2016, p. 928) are made. The authors suggest that there are similarities between perceptive and value-based decision making as explained by bounded accumulation samples of evidence. In addition, they suggest that unlike in perceptual decision making, value-based decision making accumulates the evidence required from memory.
2. The role of memory on value based decisions. The function of memory is described as the building on past events to make predictions about possible future outcomes. The authors argue that memory is important for those value-based decisions that rely on integration of information across distinct past events (sequential decision making) and also for novel choice options.
3. In the last part of the article, the authors describe the neural processes and brain areas that could be involved.

MY NOTES:

1. The idea of ‘potential value’ as possible reward system.
2. Involve memory mechanisms in the hippocampus and the ventromedial prefrontal cortex. Suggest that hippocampus contributes to value-based decisions as well as for predicting outcomes.
3. Prospection = the generation and evaluation of mental representations of possible futures. This ability fundamentally shapes human cognition, emotion, and motivation, and yet remains an understudied field of research, according to some psychologists.
4. Momentary evidence = identity of the sequence of samples of value.

Shadlen, M. N., & Shohamy, D. (2016). Decision Making and Sequential Sampling from Memory. Neuron, 90(June 1), 927–939.