Participants were eight typically developing adults, aged 19–44 years (M = 24), enrolled in undergraduate psychology courses who received extra credit for their participation. All sessions were conducted in a campus laboratory. Each participant sat at a table in front of a touchscreen computer while the primary investigator and a secondary observer sat approximately 0.5 m behind and next to the participant, respectively. Participants were presented with stimuli via Microsoft’s Visual Basic® on the touchscreen computer. Pretraining and experimental stimuli were identical to Sordello et al. (2024; see Table 1). Participants received written instructions for each condition (Hanson et al., 2022; see Table 2). The university’s institutional review board approved all procedures and recruitment protocols.

The primary dependent variable was the percentage of correct go and no-go trials across conditions, defined as correct and unprompted responses (i.e., touching related sample and comparison combinations and refraining from touching unrelated combinations; Lantaya et al., 2018). Secondary dependent variables included the percentage of unique and experimenter-defined tacts during tact tests and the frequency of consistent, inconsistent, and irrelevant vocal-verbal statements (Chastain et al., 2022; Sordello et al., 2024) during the protocol analysis and intraverbal test (see Table 3 for secondary dependent variable definitions). Additionally, data for trials to criterion during baseline training and reaction times across conditions were collected. Reaction time was recorded for go-trials following the presentation of the white box and either the participant’s response or the end of the trial (i.e., refraining from touching the white box resulted in a reaction time of 8 s).

Conditions were presented in the following order: pretraining, symmetry (BA/CA) pretest, transitivity/equivalence (BC/CB) pretest, baseline training and test (AB/AC), symmetry posttest, transitivity/equivalence posttest, protocol-analysis training, transitivity/equivalence posttest with protocol analysis, tact test, and intraverbal test. Pretraining, symmetry and transitivity/equivalence pretests, and baseline training and testing were identical to Sordello et al. (2024; see Table 4 for summary of conditions). The current study differed from Sordello et al. by first requiring participants to complete transitivity/equivalence posttests before protocol-analysis training. Then, participants completed an additional block of the transitivity/equivalence posttest with the talk-aloud requirement. Moreover, participants completed tact and intraverbal tests following emergence posttests (see Fig. 1 for procedural differences).

Procedural differences. Note. PA = protocol analysis

During training we utilized a one-to-many (OTM) format (i.e., participants were taught AB and AC relations; see Table 5). At trial onset, participants sat before a touchscreen computer in which the computer speakers played an auditory sample for approximately 3 s. After which a green box measuring 4.5 cm x 4.5 cm appeared at a 0-s delay. After touching it, the green box disappeared and an auditory comparison was played (0-s delay), followed by the appearance of a white box with a black outline measuring 4.5 cm x 4.5 cm, which remained on the screen for 8 s. Participants either touched (go) the box before the 8-s response interval elapsed or refrained from touching (no-go) the box for 8 s, depending on the relation between sample and comparison. When participants touched the white box, they were also instructed to say, “Click,” to provide an auditory indication to the experimenters that they were touching the box (Sordello et al., 2024). Touching related sample and comparison combinations during baseline training was followed by a tone from the computer speaker presented simultaneously with 10 points centered at the top of the monitor. No consequences or prompts were provided for responding correctly during no-go trials. Moreover, responding incorrectly during go and no-go trials produced no feedback. A 2 s inter-trial interval separated all trials and the white box remained in the window for a duration of 8 s.

The study employed a nonconcurrent multiple baseline design across participant dyads (Watson & Workman, 1981). This design served to rule out the possibility that repeated exposure to stimuli would lead to stimulus-class formation prior to training. Participants in the first tier completed one block of symmetry and transitivity/equivalence pretests, whereas participants in the second tier completed two blocks of each pretest. Participants within the second tier were limited to two blocks to prevent the establishment of spurious stimulus relation(s), as well as fatigue.

Interobserver agreement (IOA) was assessed across all trials and conditions between the primary investigator and the computer program and across 95.5% of trials and conditions between the primary investigator and a secondary observer. Although the computer program recorded correct and incorrect go/no-go responses, a secondary observer was used in the event of a computer malfunction and to record responses during the protocol analysis, tact, and intraverbal tests. We calculated IOA by dividing the number of agreements by the sum of agreements and disagreements and multiplying by 100. Interobserver agreement averaged 100% between the primary investigatory and the computer, 99.7% (range, 99.4-100%) between the primary and secondary observers during S-MTS, tact, and intraverbal tests, and 99.3% (range, 97.2–100%) for statements during the protocol analysis.

Procedural fidelity (PF) was recorded by the secondary observer during all baseline-training trials for correct timing of the primary investigator’s prompt delivery during go trials (i.e., 4 s), and prompts only occurring during go trials. We calculated PF by dividing the number of correctly implemented baseline trials by the total number of baseline trials and multiplying by 100. Procedural fidelity averaged 99.8% (range, 95.8–100%).