Understanding how learners form categories—and how our assessment methods might unintentionally affect their performance—has real implications for clinical practice. This study examines whether requiring learners to “think out loud” during testing changes what they appear to know, and whether verbal behavior plays a role in auditory categorization tasks. The findings offer practical guidance for clinicians designing probes and interpreting results.
What is the research question being asked and why does it matter?
The study asked two practical questions about teaching sound “categories” using a go/no-go, one-at-a-time matching task (successive matching-to-sample). First, do people still show new, untrained “go together” skills (equivalence) when the sounds are nonverbal, everyday noises—like a toaster popping or a sliding door? This matters because many learners struggle with tasks requiring them to scan several choices at once. Clinicians often want ways to teach listening and sorting skills without heavy visual demands.
Second, does making someone talk out loud during testing change their performance? Clinicians sometimes ask learners to explain their thinking (“Tell me why you chose that”), and that can either help or hurt. If talking out loud makes the task harder, you might get an unfair picture of what the learner actually knows.
A third, related question was whether verbal behavior might be part of how people solve these tasks. Are participants quietly naming sounds to themselves and using those names to get correct answers? If verbal mediation is doing most of the work, the same teaching method may not work the same way for learners with limited speech, weak echoics, or weak tacting of sounds.
What did the researchers do to answer that question?
Eight college students sat at a touchscreen computer and learned to respond “go” (touch a box) when two sounds “went together,” and “no-go” (do not touch) when they did not. Sounds were presented one after the other in the same spot, not in an array. During training, correct “go” responses earned points, but “no-go” trials did not—even when correct. Students were trained on A–B and A–C relations (one-to-many), then tested for untrained relations like B–C and C–B (transitivity/equivalence) without feedback.
The key twist was timing of the talk-aloud step. In earlier work, participants had to talk out loud during initial posttests, and fewer people passed. Here, participants first took the equivalence posttest in silence. Only after that did they get brief training on how to “think out loud,” then took one more block of the equivalence posttest while talking.
After the matching tests, participants completed two short vocal tests. In a tact test, they heard each sound and had to say what it was. In an intraverbal test, they heard one sound and had to say two other sounds that “went with it.” The researchers compared what people said during talk-aloud and these tests to their accuracy on the matching task.
How you can use this in your day-to-day clinical practice
If you are considering go/no-go successive matching (one sound, then the next, then respond), this study supports it as a possible teaching format for auditory categorization—at least with verbally capable adults. Six of eight participants showed untrained B–C/C–B relations after being taught A–B and A–C, which is the basic pattern clinicians look for when they want category-like responding to emerge.
That said, this is not an applied study. It does not show that the same outcomes will happen with children, autistic learners, or learners with limited language. Use it as a possible tool, not a promise.
Be careful about when you ask the learner to explain their thinking. Earlier research suggested that talk-aloud instructions during initial testing may lower the chance of showing emergence. Here, participants tested first without the talk-aloud requirement, and the overall pass rate was higher than in the prior study that required talk-aloud earlier.
A practical takeaway: when probing what a learner already knows, avoid adding extra response demands that were not part of training. If you suddenly add “tell me what you’re thinking” during the probe, you may be testing a different skill—dual tasking, coping with social pressure, or speaking under time limits—not the category skill itself.
If you do want verbal report data for supervision, problem-solving, or learner insight, consider collecting it after you have confirmed performance without extra demands. A clinical sequence that matches this logic: first run a clean probe with the same motor response as training, then run a second probe where the learner is invited to talk, gesture, or show their strategy. Treat the talk-aloud probe as extra information, not the main pass/fail decision. This protects the learner from being marked “wrong” just because talking made the task harder.
Do not assume that talking out loud helps. For one participant, performance got worse during the talk-aloud block. Expect that some learners will do fine, some will do better, and some will do worse when required to speak.
This is especially relevant for learners with slow speech, apraxia, stuttering, anxiety, selective mutism, or a history of being corrected for how they talk. For those learners, demanding speech during probes can be a dignity issue as well as a measurement problem. You can still offer optional language (“You can tell me your plan if you want”) instead of requiring it.
This study also suggests that verbal mediation may track with correct responding, but it does not prove that verbal behavior caused the correct performance. The researchers found that participants’ class-consistent statements matched their correct go/no-go responses very closely.
Clinically, you can treat this as a hint: if a learner has strong echoic and tact repertoires, they might use self-talk or naming to support sorting and listener tasks. Plan teaching that allows, but does not force, that strategy. If the learner naturally labels sounds or repeats them, you might not need to block that. If the learner does not label sounds yet, do not assume equivalence will emerge the same way.
Use the tact and intraverbal idea as assessment tools, not requirements. After teaching sound-to-sound or sound-to-picture relations, probe whether the learner can label the sounds (tacts) and say what goes with what (intraverbals). If matching is strong but tacting is weak, you learned something useful: the learner may know the relations without having stable names for the sounds. If tacting is strong but matching is weak, you may need more discrimination training or better observing responses.
Watch the reinforcement arrangement if you adapt this procedure. Correct “go” trials earned points, but correct “no-go” trials did not. In practice, that could bias some learners toward pressing (or toward waiting) depending on their reinforcement history and tolerance for ambiguity.
If you see a learner who “always touches” or “never touches,” do not jump to “they don’t get it.” First consider whether your consequences, trial counts, and response effort are pushing a bias. You may need clearer reinforcement for correct inhibition or other bias-reduction strategies—while keeping the task respectful and not overly punitive.
Finally, stay within what the data can support. This was a small study with college students in a controlled setting, using specific training criteria and sound sets. It suggests that talk-aloud requirements can sometimes interfere with initial emergence tests and that verbal behavior often lines up with performance.
It does not tell you the best way to teach equivalence to clinical learners, how durable the skills are, or whether these methods improve real-life listening and communication. Use these findings to guide how you probe and interpret errors. Keep your clinical decisions tied to the learner’s needs, communication preferences, and day-to-day goals.
Works Cited
Dingus, C. S., Hanson, R. J., Miguel, C. F., Stern, S., & Brand, D. (2025). Verbal mediation during auditory equivalence class formation using go/no-go successive matching-to-sample. *The Analysis of Verbal Behavior, 41*(11), 11–25. https://doi.org/10.1007/s40616-024-00209-3
