Document Type



In an early experiment, Greenspoon (1955) confirmed that adult
human verbal behavior may come under the control of subtle
differential reinforcement contingencies. Later, Truax (1966)
demonstrated that differential reinforcement played a role in therapistpatient
interactions during "non-directive" (Rogerian) therapy. In a
laboratory context, Catania, Matthews, and Shimoff (1982) found that
selectively reinforcing rules in opposition to contingencies resulted in
partiCipants performing at rates that minimized acquisition of momentby-
moment monetary reinforcement. Ninness, Shore, and Ninness
(1999) suggested that differential reinforcement of rule selections is a
powerful procedure during human-computer interactions. In this study,
we explored the possibility that similar procedures might be obtained
using voice-interactive software. During baseline, 4 subjects made
voice selections of items displayed on the screen and were exposed
to "chimes" as a form of auditory feedback when making selections. In
the first differential reinforcement condition, subjects were asked to
make voice selections of these same items and participants received
brief, positively inflected, comments from the computer (e.g., "mmmhmmm,"
"okay;' ''yes'') when making particular types of verbal
selections. Selection of other items resulted in similar comments from
the computer; however, these comments were devoid of "positive"
inflection. In the final session, contingencies were reversed, and
differential response rates regressed toward baseline levels. Results
suggest that verbal selections of particular items come under the
influence of positively inflected, computer-generated, differential
reinforcement procedures.