|
Selected Topics in Theoretical, Philosophical and Conceptual Issues |
Saturday, May 26, 2012 |
1:00 PM–1:50 PM |
605 (Convention Center) |
Area: TPC |
Chair: Denis P. O'Hora (National University of Ireland, Galway) |
|
Continuous Response Dynamics in the Experimental Analysis of Behavior |
Domain: Basic Research |
DENIS P. O'HORA (National University of Ireland, Galway), Rick Dale (University of Memphis) |
|
Abstract: In recent years, technological advances have allowed researchers to begin to investigate features of the dynamics of behavior during a response. Researchers have examined eye movements, computer mouse movement and other continuous indices of behavior. This work has primarily been done with human participants but it echoes previous work by behavior analysts on animals. The current paper reviews literature on continuous response dynamics in cognitive science in order to identify opportunities for basic behavioral research. |
|
A Neural Network Analysis of Diversified Physiological and Behavioral Outcomes |
Domain: Applied Research |
CHRIS NINNESS (Stephen F. Austin State University), Robin Rumph (Stephen F. Austin State University), Logan Clary (Stephen F. Austin State University), Judy Lauter (Stephen F. Austin State University), Michael Coffee (Stephen F. Austin State University), Sharon Ninness (Angelina College), Elizabeth Kelly (Stephen F. Austin State University), Marilyn Rumph (Stephen F. Austin State University) |
|
Abstract: Over the past 20 years a new and infinitely more robust form of data analysis has been developed, and it is only now becoming an alternative methodology of choice among behavioral researchers who are familiar with the process. Neural networks finally have surfaced because of the escalating need to make sense of the barrage of new data and data types. One of the primary attributes of neural networks is, pattern recognition, the ability to aggregate, organize, and classify an astonishingly large number of diversified data types quickly, reliably, and accurately. At a time when the behavioral/scientific community is being inundated with an ever increasing highly diversified array of new data types, the data itself is quickly becoming part of the scientific problem. With the continual influx of new data and new data types, artificial neural networks are playing a critical role in disambiguating the influx of raw diversified behavioral and physiological data. The paramount feature of our newest version of the psyNet SOM artificial neural network is its ability to organize and find common behavioral patterns among extremely large, extraordinarily dissimilar, data types and datasets. We tested our version of the SOM neural network with three behavioral and neurophysiological datasets in an effort to obtain clear output patterns that would not be identifiable by way of visual inspection of the raw data or by employing any form of traditional statistical methodology. From our perspective, artificial neural networks are going to play a critical role in disambiguating the influx of raw diverse data being brought to the attention of behavioral researchers. We will provide evidence that behavior analysts will benefit from becoming more familiar with the growing possibilities provided by various types of artificial neural network procedures. Particular emphasis will be placed on the benefits of employing our version of the self-organizing map (SOM) neural network in a wide range of behavior analytic research endeavors. |
|
|
|