act2probrat: Convert output activation to a rating of outcome probability

In catlearn: Formal Psychological Models of Categorization and Learning

Convert output activation to a rating of outcome probability

Description

Logistic function to convert output activations to rating of outcome probability (see e.g. Gluck & Bower, 1988).

Usage

 act2probrat(act, theta, beta) 

Arguments

act	Vector of output activations
theta	Scaling constant
beta	Bias constant

Details

The contents of this help file are relatively brief; a more extensive tutorial on using act2probrat can be found in Spicer et al. (n.d.).

The function takes the output activation of a learning model (e.g. slpRW), and converts it into a rating of the subjective probability that the outcome will occur. It does this separately for each activation in the vector act. It uses a logistic function to do this conversion (see e.g. Gluck & Bower, 1988, Equation 7). This function can produce a variety of monotonic mappings from activation to probability rating, determined by the value set for the two constants:

theta is a scaling constant; as its value rises, the function relating activation to rating becomes less linear and at high values approximates a step function.

beta is a bias parameter; it is the value of the output activation that results in an output rating of P = 0.5. For example, if you wish an output activation of 0.4 to produce a rated probability of 0.5, set beta to 0.4.

Value

Returns a vector of probability ratings.

Note

As this function returns probabilities, the numbers returned are always in the range 0-1. If the data you are fitting use a different range, convert them. For example, if your data are ratings on a 0-10 scale, divide them by 10. If your data are something other than probability estimates (e.g. you asked participants to use negative ratings to indicate preventative relationships), don't use this function unless you are sure it is doing what you intend.

Author(s)

Andy Wills

References

Gluck, M.A. & Bower, G.H. (1988). From conditioning to category learning: An adaptive network model. Journal of Experimental Psychology: General, 117, 227-247.

Spicer, S., Jones, P.M., Inkster, A.B., Edmunds, C.E.R. & Wills, A.J. (n.d.). Progress in learning theory through distributed collaboration: Concepts, tools, and examples. Manuscript in preparation.

catlearn documentation built on April 4, 2023, 5:12 p.m.