Confidence-Related Decision Making

Abstract

Neurons have been recorded that reflect in their firing rates the confidence in a decision. Here we show how this could arise as an emergent property in an integrate-and-fire attractor network model of decision making. The attractor network has populations of neurons that respond to each of the possible choices, each biased by the evidence for that choice, and there is competition between the attractor states until one population wins the competition and finishes with high firing that represents the decision. Noise resulting from the random spiking times of individual neurons makes the decision making probabilistic. We also show that a second attractor network can make decisions based on the confidence in the first decision. This system is supported by and accounts for neuronal responses recorded during decision making and makes predictions about the neuronal activity that will be found when a decision is made about whether to stay with a first decision or to abort the trial and start again. The research shows how monitoring can be performed in the brain and this has many implications for understanding cognitive functioning.

On the Relation between the Mind and the Brain: A Neuroscience Perspective

“…Another is to consider the process as a mechanism with different levels of explanation, in the following way. We can now understand brain processing from the level of ion channels in neurons, through neuronal biophysics, to neuronal firing, through the computations performed by populations of neurons, and how their activity is reflected by functional neuroimaging, to behavioural and cognitive effects [Rolls 2008b, Rolls & Deco 2010, Rolls 2012c]. Activity at any one level can be used to understand activity at the next.…”

“…These functions include many aspects of perception including visual face and object recognition, and taste, olfactory and related processing; short-term memory; long-term memory; attention; emotion; and decision-making [Rolls 2008b], [Rolls & Deco 2010], [Rolls 2012c[Rolls , 2014[Rolls , 2010a[Rolls , 2012b. Predictions made at one level can be tested at another.…”

“…Sometimes the cognitive effects seem remarkable, for example the recall of a whole memory from a part of it, and we describe this as an 'emergent property', but once understood from the mechanistic level upwards, the functions implemented are elegant and wonderful, but understandable and not magical or poorly understood [Rolls 2008b], [Rolls & Deco 2010], [Rolls 2012c]. Different philosophers may choose or not to say that causality operates between these different levels of explanation, but the point I make is that however they speak about causality in such a mechanistic system with interesting 'emergent' computational properties, the system is now well-defined, is no longer mysterious or magical, and we have now from a combination of neuroscience and analyses of the type used in theoretical physics a clear understanding of the properties of neural systems and how cognition emerges from neural mechanisms.…”

Chronological Prospective of Behavioral Decision Making: A Three Realm Model for Cognitive Processing and Behavior

“…The brain then decides on the type of behaviour to be displayed based on permutations and combinations of these coded information for a particular situation in both reward and punishment circuits and finally giving way to the stronger signal. This hypothesis finds support from the 'integrate-and-fire attractor network model' which has been electro-physiologically tested and elaborately discussed by Insabato et al [18]. The model suggests that there are populations of neurons in an attractor network which respond to each of the possible choices, biased by evidence for the choice.…”

“…Similarly, Schacter et al [51] in their constructive episodic simulation hypothesis propose that episodic memory supports the construction of predictive future events. Szpunar and Schacter further observed that increased plausibility of the future event was associated with rewarding (positive) or punishing (negative) emotional events [52]. Based on the integrateand-fire-attractor network model of [53] it may very well be proposed that neural networks of both the reward and punishment realms related to an event may be simultaneously activated and the network generating stronger potentials determines whether the ensuing behaviour is to obtain a reward or avert a punishment.…”