False-Positive Tangible Outcomes of Functional Analyses

Functional analysis (FA) is a precise method for identifying variables maintaining problem behavior. However, results can be influenced by sensitivities to assessment conditions, potentially leading to false positives, especially with tangible reinforcement.

Study Overview

This paper examines tangible reinforcement susceptibility by:

Determining if a new response is acquired more readily with a tangible contingency compared to others in an FA (Study 1).
Determining if problem behavior, not socially maintained, emerges with tangible reinforcement (Study 2).

The results suggest caution when including tangible items in FAs, basing selections on items typically in the individual’s environment.

General FA methodology

FA involves systematically varying antecedent and consequent events to observe their effects on behavior. Despite its precision, idiosyncratic sensitivities can influence results.

Tangible Reinforcers

McCord and Neef (2005) discuss that tangible items in the attention condition of an FA may mask the effects of attention and produce a false-negative outcome. The study focuses on tangible items producing a false-positive outcome, incorrectly identifying a tangible function.

Multiple Control

Beavers and Iwata (2011) examined FA studies and found that tangible reinforcement was reported as one source of maintenance in 41 cases. Exposure to preferred stimuli may have produced a false-positive outcome in some cases.

Direct Evidence

Shirley, Iwata, and Kahng (1999) found high levels of mouthing during the tangible condition of an FA. A DA showed the only tangible consequence was a towel, and a repeat of the FA with the towel showed identical levels of mouthing in both conditions. Thus, tangible items selected from a preference assessment produced high rates of hand mouthing, whereas the tangible consequence identified through naturalistic observation had no effect.

Galiatsatos and Graff (2003) conducted an FA of one boy’s screaming behavior and observed that screaming occurred most often in two tangible conditions: access to a toy and access to an edible item. They then conducted a DA in the student’s classroom but never observed the delivery of any tangible item as a consequence for screaming. Tangible items, especially edible items, are highly effective reinforcers that tipically enhance their value and are likely to produce new behavior, and also to increase the occurrence of behavior already in one’s repertoire. The present studies explored both possibilities.

Study 1: Acquisition of Behavior Under Tangible Reinforcement

Examined whether a new response is more readily acquired with tangible reinforcement vs. other FA contingencies, to see if behavior is more susceptible to tangible reinforcement.

Method

Six participants with and without disabilities. An arbitrary task (switch pressing or card touching) was selected as the dependent variable for each subject. Preference assessments identified preferred edible and leisure items.

Functional Analysis Conditions

Alone, attention, demand, play, and tangible conditions were alternated. The tangible condition involved delivering a bite-size portion of a highly preferred edible item after the target response. The switch or card was located near the source of reinforcement.

Results

Milton and Rusty's responding emerged over several exposures to the tangible condition, whereas Bobby, Jane, and Evan immediately began responding in the tangible condition. All five subjects exhibited very little responding in any other condition. By contrast, Justin engaged in high rates of responding only in the alone condition. Data for five of the six subjects shows that responding is more susceptible to tangible reinforcement. Justin was the only subject who did not acquire the target response in the tangible condition. Thus, some aspect of switch pressing per se was sufficient to maintain responding.

Study 2: False-Positive Tangible Outcomes During Functional Analyses of Problem Behavior

Explored whether problem behavior, maintained by automatic reinforcement, would be affected by tangible contingencies.

Method

Three participants with stereotypy. Descriptive assessment (DA) identified consequences following stereotypy in the natural environment. The tangible item was selected as the tangible DA item and was used in the comparison of tangible conditions. Functional analysis included alone, attention, demand, and play conditions.

Comparison of tangible conditions.

Three conditions were conducted in a multielement design:

alone condition of the FA
tangible preference assessment condition, in which the most highly preferred edible item from the paired-stimulus preference assessment was delivered following each instance of stereotypy
tangible DA condition, in which the stimulus most frequently delivered during the DA was delivered after each instance of stereotypy

Results

Amy's and Oliver's stereotypy occurred more often and consistently in the tangible preference assessment condition. Jimmy initially engaged in somewhat less stereotypy in the tangible preference assessment condition. All subjects’ FA results indicated that stereotypy was maintained at least in part by automatic reinforcement.

Discussion

A response with no history of reinforcement is highly susceptible to tangible reinforcement. Stereotypy maintained by automatic reinforcement was more likely to occur under tangible reinforcement.

Recommendations

Avoid tangible FA conditions unless there's a strong suspicion of tangible reinforcement. When included, tangible conditions should reflect items actually delivered for problem behavior.

Qualifications

Additional functions may have emerged with longer condition durations. Also, the inclusion of edible items in both studies makes it unclear if the effects observed also would have been obtained with leisure items. Finally, data from Study 2 may have limited generality to problem behavior other than that maintained by automatic reinforcement.

Clarification of results required further analysis. The fact that such findings and the subsequent need to make procedural adjustments have been reported rarely underscores the general robustness of the conditions contained in a typical FA.