Note: This article was originally published in the October 2015 issue of Behavior Analysis Quarterly, Vol. (1)3.
By Ben Witts, PhD, BCBA
St. Cloud State University
I have a fun exercise I conduct with my incoming master’s students (well, fun for me, anyway). I ask them to list several discriminative stimuli (SD; pronounced es-dee) and I write them on the marker board. For example, students might say that a keyboard is the SD for typing, or that a headache is the SD for getting aspirin. Then, we systematically eliminate roughly 80% of the examples as they are not truly SDs. While the exercise helps students to realize that they’re in the right place—as they’ve got a lot to learn—it’s a bit disturbing to realize that so many students, many of whom have been practicing behavior analysts for years, don’t know how to identify a discriminative stimulus.
Misconceptions about SDs
When I survey my students, I get a typical “definition” of discriminative stimuli that equates to something like, “A discriminative stimulus is a stimulus that occasions a response.” Here ‘occasion’ refers to a cause. So the belief is that a discriminative stimulus causes a response. Were this to be true, then discriminative stimuli would better fit in a respondent analysis. Recall that respondents, and their reflexive base, stand in a close relation where often a particular stimulus causes a particular response (though that analysis is, perhaps, simplistic).
Some students will counter that an SD is really just a stimulus that makes a response more likely. This definition is a bit closer to what an SD really is, but it lacks the necessary features of discriminative stimuli. Here the students tend to confuse prompts and motivating operants with SDs. Prompts often refer to antecedent stimuli that help to promote a weak, incomplete, or low-probability response to production. Prompts are used when an organism needs a bit of help in producing the correct response. Motivating operations alter reinforcement effectiveness, giving rise to responding that produces such reinforcement.
Of course, my sample is biased here as I only turn to my students. However, given that definitional errors similar to what I’ve pointed out here appear in popular texts, peer-reviewed journal articles, and other outlets like blogs, discussion boards, and Facebook posting exchanges, I suspect they are rather common. So what, then, is a discriminative stimulus?
Defining the Discriminative Stimulus
Jack Michael (2004; pp. 59-65) provided a clear definition of discriminative stimuli, and it’s the one I use here. A discriminative stimulus is a stimulus that, when present, results in an increase in the rate of a particular response (or class). While this sounds like the prompt described above, some caveats come with the SD.
First, the SD gains its status as the product of a history of differential reinforcement. Here responding is reinforced in its presence, but not in its absence. When responding occurs in its absence, an extinction schedule is in effect. When present, stimuli that signal the differential unavailability of reinforcement are call SΔs (es-delta). For example, while typing this, my Internet connection has gone out. I know this because an icon with an exclamation mark in a small yellow triangle on my status bar lets me know that attempts to access the Internet will not be met with reinforcement. In this case, the icon is an SΔ for accessing the Internet. When the icon is removed and I’m left with a series of bars indicating access to the Internet has returned, my responses with respect to the Internet will be reinforced (thus, these bars are an SD for accessing the Internet).
Second, the SD is only effective under appropriate motivating operations. This second caveat is important, as this requires that the organism would respond in the absence of the stimulus if reinforcement were likely. Thus, the SD signals the differential availability of a particular reinforcer given the response and related motivating operation. In the example above with the Internet connection, I find that I have need to access the Internet to check on important emails. Thus, responding with respect to the Internet is highly probable in both cases. However, if I was focused on writing this article, access to the Internet might be distracting and I would find its use of no reinforcing value. In this latter case, I care for neither icon, as I would not find need to use the Internet (and thus neither has discriminative properties at that time).
Inherent in this definition is a third caveat; the SD does not differentially restrict responding (cf. Malott, 2008; pp. 217-218). Keeping with the Internet connection example, the icons do not restrict my attempts to check email (the response in question). Instead, the icons have unique histories of reinforcement with respect to the response in question and its related reinforcement. In other words, I can check my email all I want; it’s only under some conditions that this response is reinforced.
This might seem like a lot to consider, but it’s actually quite simple to determine if a stimulus is a discriminative stimulus. I’ve provided a quick yes/no diagram to help you determine SD status.
Now that you know just what a discriminative stimulus is, don’t be surprised if you find professors, colleagues, and even authors in peer-reviewed journal articles calling all manner of antecedent stimuli “SD” when they meant something else.