The Effects of Vocal Blocking on Sequencing Visual and Tactile Stimuli

Abstract

The inclusion of private events in the philosophy of our science is integral to avoid dualism and remain objective rather than making assumptions about an unseen mind. However, the inclusion of behaviors and stimuli which cannot be observed in an analysis poses obvious issues. One established method of studying covert behavior is to examine tasks that are presumed to require verbal mediation, and observing how a participant’s performance is affected when they are required to speak out loud during the task (often called “blocking”), again presuming this will make it difficult or impossible to simultaneously talk to yourself covertly. This study investigated the effects of vocal blocking on a sequencing task, or lining things up in a specified order. In one experiment, the items sequenced were abstract line drawings, and a second experiment used differently textured fabric stimuli (or “tactile cards”). In the second experiment, participants learned to tact and then sequence the tactile stimuli while they were blindfolded. The effect of vocal blocking on putative covert rehearsal was dissimilar across the two modalities of the experiments. This preliminary study provides insight into the nature of covert behavior as it relates to different senses and opens questions about the generality of studies examining covert mediation.

Behavior analysis as a science has focused primarily on the empirical study of overt (observable) behaviors and their relationship to environmental variables. However, radical behaviorism encourages us to view all behavior, observable or not, as following the same principles and participating in a unified science of behavior. Many behavior analysts consider this to be one of the (if not the) philosophical distinctions between Skinner and Watson’s earlier methodological behaviorism (Baum, 2017, pp. 9–10; cf. Moore, 2008, pp. 214–215). Private events are any element of an individual’s behavioral event which cannot be observed by another person. Skinner typically used the term “private” to refer to stimuli within the skin (e.g., 1953, Ch 17; 1957, pp. 130–141 & throughout) and “covert” to refer to unobservable behavior (e.g., 1953, pp. 263–264; 1957, throughout). Our behavior analytic verbal community, however, has not upheld the distinction, using “private events” to refer to stimuli or behavior (explicitly defined by Palmer, 2020, p. xi) and “covert” when referring to responses, either generally or specifically.

There is a great deal of complex human behavior which arguably requires the inclusion of private events to be understood without cognitive explanatory fictions (Palmer, 2009). Even an exchange as ubiquitous as answering “what are you doing?” with “I was thinking about ____” demands a behavior analytic account, the only alternative being mentalism. The radical behaviorist’s answer has been to work out the principles of behavior with observable phenomena, and then assume that the unobserved carry no special qualities. Indeed, this is the method by which interpretation is used in all natural sciences (Palmer, 2011). But interpretation is not a stopping point; on the contrary, it can tell us where (or how) we might look to add empirical evidence to our explanations. To that end, behavior analysts are developing techniques to study covert behavior through the observation of overt behaviors measured under different conditions which may reveal characteristics of covert behavior.

Past research has studied mediating responses, which are public or private, verbal or nonverbal behaviors that can include physical movement, repeating a list of words to oneself, or naming objects to improve the accuracy of recall (e.g., Blough, 1959; Lowenkron, 1984; Torgrud & Holborn, 1989; Sundberg et al. 2018). An early seminal paper in this area was “Delayed Matching in the Pigeon” by Blough (1959). The author sought to investigate the duration of stimulus control, and how behavior might be controlled by stimuli once those stimuli were no longer present. In this experiment, pigeons were exposed to a flickering or steady lit sample for one second, and immediately upon the offset of the sample two key peck targets were turned on, one flickering and one steady. Key pecks on the targets which matched the sample were reinforced, and non-matching key pecks were extinguished. Once this immediate matching-to-sample task was learned, delays between the sample offset and comparison target onsets were introduced. While two of the four participants’ performance degraded quickly, down to near chance at delays of 5 s, the other two were able to maintain accurate responding at 5 and even 10 s. These latter two birds were observed to be engaging in repetitive, seemingly superstitious responses during the delays. More importantly, these stereotyped responses had different topographies for the two stimuli. Blough called these responses “mediating” and concluded that once they were learned, the sample stimulus evoked a particular mediating response, and then once the comparison keys were lit up, the stimuli generated by mediating responses could control pecking the correct key.

This appeal to mediation as an explanation for delayed stimulus control is very attractive, considering how common an experience it is for a person to repeat something in order to remember it. Additionally, since behaviorists view overt and covert verbal behavior as having little (or no) difference aside from observability, covert mediation is an appealing explanation for some instances of delayed matching-to-sample (or similar delayed stimulus control) in people. To demonstrate covert mediation is occurring is obviously not straightforward, since it cannot be observed by definition.

Many studies since Blough (1959) have designed delayed matching-to-sample or complex sequencing tasks involving unfamiliar stimuli while incorporating blocked trials, or trials which require participants to engage in unrelated overt verbal behavior intended to prevent the participant from engaging in covert verbal behavior. The logic of this technique is that one cannot engage in overt and covert verbal behavior simultaneously. Many of these studies fall under the concept of “joint control” an explanation for behavior where one topography is simultaneously evoked by two separate stimuli, and the resulting jump in response strength is a discriminable event (see Ampuero & Miklos, 2019 for a relevant review). One study required participants to place four picture cards of common objects under the control of corresponding Chinese Mandarin words in specified sequences (Gutierrez, 2006). After learning Mandarin tacts for pictures, researchers said a sequence of four words and tasked participants with putting the pictures in that order. During baseline trials, participants were free to repeat (rehearse/mediate) the tacts overtly or covertly to help complete the correct sequence. The “joint control” was met when the sequence they were rehearsing matched the tact of the picture cards in order. During “blocked” trials, participants were instructed to place picture cards in sequence while singing “The Wheels on the Bus,” assumed to be incompatible with covert (and certainly overt) verbal mediation. Gutierrez found that the number of correct sequences declined when mediating responses were blocked by singing.

Another study (Clough et al., 2016, Experiment 1, 3) followed a similar design by requiring participants to sequence abstract picture cards after learning corresponding nonsense names for the figures. This study similarly explained the terminal performance in terms of joint control: when provided the nonsense names for a sequence of pictures, participants could echo that sequence and repeat it while arranging the pictures; then, when the arrangement of pictures evoked (as a tact) the same sounds they were echoing, participants knew (by discriminating the jump in response strength) that the task was complete. Clough and colleagues found that the percentage of correct responses greatly declined in trials where participants were required to engage in vocal blocking (singing “Happy Birthday” while sequencing). If the participants couldn’t rehearse the sequence there was no echoic to enter into joint control with any arrangement of picture sequence tacts. Contrastingly, the results of a study which used a visual–visual delayed matching-to-sample task with young children suggested that verbal rehearsal may not be required for a correct response to be emitted (Ratkos et al., 2016). Clough et al. (Experiment 2) and Lowenkron (1988) used hand signs instead of vocal tacts to the same effect, mediating by maintaining this physical “rehearsal” could enter into joint control the same way a vocal echoic could.

Joint control research has not been limited to basic experimental analyses; researchers have already begun demonstrating the applied value of teaching the responses required for joint control for valuable tool skills relevant in people’s lives. Causin et al. (2013) prompted three children with limited verbal skills aged 6–17 years to rehearse multi-item requests, e.g., “give me the ball, cup, and spoon.” After learning to rehearse (vocally and with manual signs) immediately after instruction, participants learned to select multiple picture cards in order corresponding with a request, and this capability generalized to untrained sets. Vosters and Luczynski (2020) taught children with ASD aged 5–6 years to follow multi-step instructions with delays between instruction and opportunity to respond using overt and covert rehearsal. These instructions had two parts with three varied stimuli, such as “bring me the crayon, put the grapes in the bag.” Some delays in training were contrived using a cardboard cover to keep objects out of view for 10 s and 15 s, then moving to keeping target objects in another room down a 10 m hallway. Performance generalized to novel combinations of stimuli and parent-delivered naturalistic instructions in the children’s homes.

The studies outlined in this introduction have provided insight into the role of mediating responses in both sequencing and delayed matching-to-sample tasks involving visual, vocal, and gestural stimuli. However, we cannot locate any research on the role of mediating responses when recalling tactile stimuli. Even in studies where hand signs are used (Clough et al., 2016, Experiment 2; Lowenkron, 1988), participant performance is likely a result of a tact of the visual stimulus of the hand sign (rather than proprioception) entering into joint control with the sample stimuli. It may be the case that vocal rehearsal is a requirement for mediating delays in these experimental preparations as a matter of principle and sense modality is irrelevant. However, our learning histories for different sense modalities may show different strategies or performance effects when doing equivalent tasks where the stimuli are felt rather than seen. The patterns found in previous studies could be an artifact of all joint control studies using stimuli which are seen, rather than a property of joint control, rehearsal, or the duration of stimulus control. Even with the variability of methods of previous joint control studies, as a field we may be drawing conclusions with regard to private events which have only been demonstrated under the control of visual stimuli.

The purpose of this study is to determine if stimuli experienced through different sense modalities (sight versus touch) changes the effect of vocal blocking on tasks presumed to require mediating responses. Therefore, two experiments were completed to first independently replicate a recent experiment with visual stimuli, and then repeat the experiment changing the sense modality and keeping all other methods as close to the original as possible. The first experiment is a replication of Clough et al.’s Experiment 3 (2016, pp. 258–261), a visual sequencing task, and the second experiment uses the same procedures applied to tactile stimuli with blindfolded participants. These studies will replicate and expand upon previous research in pursuit of furthering our field’s understanding of private events.

Method

General Procedures

The two experiments followed the same order of conditions. First, participants were asked to put four novel, unlearned stimuli in a sequence under control of the experimenter saying a sequence of four (again unlearned) stimulus names to show that any skill with this task demonstrated later was a result of learning during the experiment. Then, the participants learned to echo these stimulus names and to tact each stimulus. After learning to echo the labels and to tact each stimulus, the sequencing test was repeated, with alternating blocks of trials where participants could verbally (overtly or covertly) rehearse the sequence and blocks of trials where rehearsal was blocked by compelling the participants to repeatedly say the alphabet out loud while arranging the stimuli. This arrangement tested if vocal blocking impacts the sequencing task by disrupting mediation.

A possible confound is that saying the alphabet could simply be a distraction that makes the sequencing task harder without any reference to any presumed covert mediation. Therefore, at each step in the process, a block of trials where the sample sequence was present throughout the trial (as a match-to-sample) was conducted as a control.

The set of stimuli used for matching-to-sample vs sequencing tasks was alternated across participants to control for any inherent difference in difficulty sequencing the two sets. The order of stimuli sequences used in each condition were randomized across the study and then adjusted to ensure a sequence was not repeated within a single block of trials.

Participants were informed participation would take about one hour, and that they could take a break (or leave the study altogether) whenever they wished. Participation took between 48 and 68 min for all participants, none took any significant break (incidental breaks of 10–20 s occurred occasionally between blocks of trials) nor left the study.

Procedural Fidelity and Interobserver Agreement Measures

Across both experiments, the second author reviewed video recordings to measure interobserver agreement (IOA) of the dependent variables and treatment integrity of the experiment’s procedures. IOA was measured by scoring participant responses as correct or incorrect and then comparing each trial to the data collected live during the session and checking if the two measures matched. IOA was calculated by dividing the number of agreements by the total number of trials and multiplying by 100 to get a percentage. IOA was scored for 37.5% of all sessions with an average of 99.3% (range 87.5–100%) agreement.

Procedural fidelity (PF) was measured with a procedure checklist which detailed 7–13 criteria for each trial (the number of criteria were different for each phase), such as which card set was used, the order of cards presented, error correction procedures, type of consequence given (i.e., praise during training and neutral responses during testing). Across all conditions, 37.5% of sessions were scored for PF, with an average PF of 97.3% (range 86.2–100%).

Experiment 1

In developing a valid comparison between visual and tactile stimuli sequencing, an established successful protocol for sequencing visual stimuli was used. Experiment 1 is a replication of Experiment 3 from Clough et al. (2016). Not only is replication an important part of science, but conducting this experiment ourselves helped ensure that any differences in the results of our Experiment 2 were not the result of unintentional differences in preparation or execution across settings. Training and sequencing phases from Clough et al. are referred to in order to ensure precise replication. In Experiment 3, Clough and colleagues described Joint Control Training procedures but determined they were not necessary for their participants; therefore, they are not described herein.

Participants and Setting

Participants were four undergraduate students (1 male and 3 females, over the age of 18) recruited from Psychology 101 classes at a private liberal arts college. Course extra credit and a $25 gift card were provided contingent on completion of the study. Sessions were conducted in a room on campus containing a table and two chairs facing each other. The experimenter sat across from the participant at the table. Each individual participated in one session lasting approximately one hour.

Materials

Eight abstract images printed onto laminated picture cards (shown with their corresponding arbitrary names in Fig. 1) served as sample stimuli. The eight pictures were divided into two sets containing four pictures and were randomly assigned as either vocal or visual sequencing sets for each participant. For visual sequencing tests, the sequences were shown as pre-printed pictures of the stimulus sequences. To ensure procedural fidelity and increase the speed at which trials could be presented, pictures were contained within a binder. The visual stimuli and their labels were identical to the stimuli used in Clough et al. (2016).

Fig. 1.

Visual stimuli and corresponding spoken names

Dependent Measures

The primary dependent measure was the percentage of accurate sequencing responses within 5-trial blocks. The definition of accurate sequencing was placing all four picture cards in the order stated or shown by the experimenter. Orientation of the stimulus was not important for scoring a correct response; most stimuli had a distinct top and bottom, but cards could be placed upside down and be scored correct. Also, the stimuli could be placed chronologically in any order, but when the participant ended the trial, the cards had to match the order of the sample (e.g., the rightmost card could be placed first, second, etc.) for matching trials and “read” left-to-right to be correct in the vocal sequencing trials.

Correct echoics and tacts in training phases were measured to compare to criteria to continue to the next phase. Correct echoic responses matched the consonant and vowel sounds of the experimenter. Correct tacts were defined as saying the correct name of the visual stimulus in the presence of the specified visual stimulus.

Experimental Design

The experiment was conducted with an adapted alternating treatments design, moving rapidly between matching-to-sample and vocal sequencing conditions (Table 1). A reversal design was used to demonstrate differences in performance of the sequencing tasks with and without vocal blocking. This design enables us to observe differences in performance under the control of visual stimuli (matching-to-sample) and vocal stimuli (the stimulus sample sequence spoken by the experimenter once) and the difference between these tasks with and without vocal blocking.

Table 1.

Condition order and set assignment for experimental conditions in experiment 1 & 2

Set A–matching and Set B–vocal	Set A–vocal and Set B–matching
1. Sequencing test – vocal	1. Sequencing test – matching
2. Sequencing test – matching	2. Sequencing test – vocal
3. Echoic training	3. Echoic training
4. Vocal tact training	4. Vocal tact training
5. Sequencing test – vocal	5. Sequencing test – matching
6. Sequencing test – matching	6. Sequencing test – vocal
7. Vocal blocking test – vocal	10. Vocal blocking test – matching
8. Vocal blocking test – matching	11. Vocal blocking test – vocal
9. Sequencing test – vocal	12. Sequencing test – matching
10. Sequencing test – matching	13. Sequencing test – vocal
11. Vocal blocking test – vocal	14. Vocal blocking test – matching
12. Vocal blocking test – matching	15. Vocal blocking test – vocal
13. Sequencing test – vocal	16. Sequencing test – matching
14. Sequencing test – matching	17. Sequencing test – vocal

Procedures

Vocal Sequencing Tests

At the beginning of each trial, the experimenter stated, “Please attempt to put the cards in left to right order, then put your hands on your lap when you are finished, and I’m going to tell you the names of the cards in order.” The experimenter then stated the pre-arranged sequence (e.g., “nop, zeg, raz, tiv”) and then handed the participant the four picture cards in a stack, shuffled in a random order.

Following the instructions, the participant arranged the picture cards on the table in front of them, then put their hands in their lap. Nonspecific feedback in the form of “thank you” or “OK” was delivered at the end of all trials. Each vocal sequencing test phase consisted of five trials, in which five non-repeating random sequences of cards were tested.

Visual Matching Sequencing Tests

During visual matching sequencing tests, a picture of a sequence of four stimuli was shown to the participant. Then, the experimenter handed the participant a set of those stimuli to arrange in the same sequence. The experimenter told the participant, “Now I want you to attempt to put the pictures in the order you see, from left to right, and put your hands on your lap when you are finished.” The sequences of stimuli were pre-determined and semi-random, but a sequence was not repeated within a 5-trial block. For example, the experimenter said, “Please attempt to put the pictures in the order you see, from left to right, and put your hands on your lap when you are finished,” opened a binder to display a picture showing the lin, yaf, kev, and jom stimuli, and then handed the participant the lin, yaf, kev, and jom stimulus cards in a shuffled stack. The visual sample remained present throughout the trial. At the end of each trial, the experimenter responded with a neutral “thank you,” or “OK,” regardless of the accuracy of the participant’s performance, and began the next trial. Each visual matching sequencing test consisted of five trials.

Echoic Training

The participants were told, “Before we learn the names of the pictures, I just want you to repeat all their names back to me when I say it.” The experimenter then said each of the eight stimulus names and paused for 5 s between each one to allow the participant to repeat the name. Verbal praise (e.g., “that’s right,” “exactly,” “good job”, etc.) was delivered contingent on accurate echoing, and errors were followed by corrective feedback. For example, if the experimenter said, “raz,” but the participant said, “rads,” the experimenter said, “no, raz.” The participant repeated, “raz,” and the experimenter said, “that’s right,” and started the next trial. Echoic training was complete following the progression of eight consecutive training trials (one training block) without any errors.

Tact Training

Tact training was accomplished via an echoic-to-tact transfer procedure. The phase began with the experimenter stating, “Now we will learn the names of the pictures. I’ll show you each picture and tell you its name, I want you to look at the picture and repeat the name.” During this first block of trials each stimulus was placed on the table in front of the participant and the experimenter said “this is [name of stimulus],” waited 5 s for the participant to repeat the name, and delivered praise for correct responses and error correction for any errors. After a participant correctly echoed all eight stimulus names in the presence of the stimulus without errors, the immediate prompt was discontinued and participants were given 5 s to respond. The experimenter stated, “Now when I show you the picture I want you to say the name. If you’re wrong or take too long to respond, I’ll tell you the correct name.” If an incorrect response occurred, or if the participant did not respond after 5 s, an error-correction procedure was implemented in which the experimenter said, “No, this is [name of stimulus].” Typically, the participant then immediately repeated the name; if they did not, they were instructed to, “look at the card and say [name of stimulus].” Correct responses were followed by verbal praise, including correct responses during error correction. Presentation of each of the eight stimuli in a random order continued until the participant got all eight correct, twice in a row.

Vocal Blocking Tests

Vocal blocking tests required the participant to engage in continuous vocal behavior throughout the task to prevent any covert (or overt) verbal mediation. Vocal blocking tests were implemented across both vocal- and visual-matching sequence tests. All steps of the sequencing test were repeated with one exception; Participants were given the following instructions: “When I point to you, immediately begin saying the alphabet as quickly as you can, and when you get to z, start again with a.” When vocal blocking was implemented during a vocal sequencing test, the participant was instructed to begin saying the alphabet after they correctly repeated the sequence but before they were handed the stimulus cards. During visual sequencing tests, participants were instructed to begin saying the alphabet after the sample sequence was shown but before handing them the stimulus cards.

Post Experiment Interview

Participants were asked the following questions:

1. Did you use any kind of strategies to learn the name of the individual cards?

2. Did you use any kind of strategies to sequence the cards?

3. Did you talk to yourself about any of the cards?

4. Was any part of this study difficult for you?

5. Do you have anything you would like to add about your experience?

Results

Figure 2 displays the data for the percentage of accurate sequences during sequencing tests and vocal blocking tests for participants V1, V2, V3, and V4. During initial vocal sequencing tests (i.e., before learning the names of the stimuli), all participants responded at 0% accuracy. Performance during initial visual matching sequencing tests was 100% across all participants. All participants only needed one block of echoic training. Tact training required 8, 11, 11, and 7 blocks to reach mastery for the four participants, respectively.

Fig. 2.

Experiment 1 results Note. Percentage of correct sequences per each 5-Trial Block for sets using vocal procedures (closed triangles) and visual matching (open squares) for participants V1, V2, V3 and V4 across sequencing and vocal blocking conditions during Experiment 1.

All participants performed at 100% accuracy during all visual sequencing tests and blocked visual sequencing tests. During vocal sequencing tests, V2 and V3 sequenced with 100% accuracy across all three 5-trial blocks. V1 initially sequenced with 60% accuracy and then 100% accuracy in the following vocal sequencing tests. V4 performed at 80% accuracy in all three 5-trial blocks. During vocal blocking tests, all participants’ performance was reduced in comparison to their unblocked vocal sequencing tests. Participants’ sequencing accuracy was between 20% and 60% during vocal blocking tests, but performance in visual matching vocal blocking tests was unaffected by the repetition of the alphabet and remained at 100% accuracy across all participants.

During the post-experiment interview, some participants reported that when they were learning to tact the unfamiliar stimuli, they identified esoteric corresponding features between the names and their visual features. Participant V1 reported, “Kev looked like Kevin from The Office because he’s, like, really large. That sounds really weird now that I say that out loud! And like, some of them looked like they had the letters hidden within the shape. Tiv looked like there was like a V, and then like lin had like an L and an I and it looked like an N.”

The participants reported in the post-experiment interview that they repeated the trained names of the picture cards to themselves during sequencing tests. When reciting the alphabet, the participants explained that they had difficulty reciting the alphabet in the correct order and said some letters out of order. In response to the question on if they used any strategies to sequence the cards, participant V3 explained, “No, well I guess just, trying to remember, it was a lot harder when I was singing the alphabet. I guess, I felt like I could remember the first one and the last one really well, but it was harder to remember the middle ones.” Participants reported having difficulty repeating the names of these items during vocal blocking procedures, but they did not report any difficulty when visually sequencing the cards. For visual sequencing tests, the participants did not report using any specific strategies.

Discussion

The results from Experiment 1 support the results of Experiment 3 in Clough et al. (2016). Vocal blocking procedures were effective in interfering with the presumed covert verbal mediation during vocal sequencing tests, but visual matching sequencing tests were not impacted by vocal blocking. These data suggest that during visual–visual matching tasks, no mediation may be required, but mediation may be necessary for a vocal sequencing task.

Participants reported that the visual sequencing task was more difficult during blocking conditions, but they were able to correctly place cards in the correct sequence regardless. When the order of the stimuli was presented vocally, accuracy greatly decreased during blocking conditions (between 20% and 60% reduction). However, for at least one participant (V3, possibly also V4) vocal blocking reduced accuracy, but accuracy was still far higher than what would be expected via random chance. Even with a reduction in performance (100% to 60%), the accuracy of vocal sequencing during blocking shows that concurrent mediation may not be a requirement for this task.

Experiment 2

Participants and Setting

Participants were four college students (1 male and 3 females all over the age of 18) recruited from Psychology 101 classes at a private liberal arts college. None of the participants from Experiment 1 were included in Experiment 2. The contingency for study completion, setting, and duration of participation was the same as Experiment 1.

Materials

Eight tactile cards of differing textures served as sample stimuli and were assigned corresponding arbitrary names (Fig. 3). The tactile cards were made custom for this study, constructed of different fabric materials stretched over a plush rectangle, resembling very small throw pillows with flat, hard backs. Their length, width, and height measured 16 cm by 13.5 cm by 4–4.5 cm (Fig. 4). The fabrics were selected for their distinct textures (Table 2). The eight tactile cards were randomly divided into two sets of four, with the exception of A3 and B2, which the experimenters decided were too close in texture and might cause issues discriminating between them if they were in the same set. The sets were randomly assigned as the vocal or tactile matching set for each participant. The sequences for each of the five trial blocks were selected randomly but changed to ensure a sequence was not repeated in a block of trials. Video cameras were used for data collection purposes. Participants wore blindfolds, made with cloth bandanas, that prevented them from seeing the cards. A tray that could hold four tactile cards was used to easily give the blindfolded participants access to the stimuli during sequencing tests.

Fig. 3.

Experiment 2 close-up pictures of tactile stimulus cards with corresponding spoken names

Fig. 4.

Experiment 2 tactile stimulus card (nop, B4) next to a pen and sticky-note for size

Table 2.

Tactile stimulus cards with subjective descriptions of material and texture

Stimulus	Description and comparisons
A1–kev	Open-weave hard plastic grid
A2–lin	Smooth plastic, like a vinyl chair cover
A3–yaf	Soft, fuzzy cotton
A4–jom	Hairy, furry, soft; like a cat or dog
B1–tiv	Rough, hard plastic; like sandpaper
B2–zeg	Soft, smooth, like felt or fake suede
B3–raz	Rough, fine net texture over cloth
B4–nop	Soft, terry cloth, like a washcloth

Dependent Measures

Dependent measures were identical to Experiment 1; the accuracy of sequencing responses was the primary dependent variable; correct echoics and tacts in training phases were measured to compare to criteria to continue to the next phase. Orientation of the stimulus was not important for scoring a correct response, the tactile cards did not have a clear top or bottom, but were slightly rectangular.

Experimental Design

As in Experiment 1, an adapted alternating treatments design was used to alternate between sessions of tactile matching and vocal sequencing. A reversal design was also used, introducing and removing the independent variable of vocal blocking.

Procedures

Changes from Experiment 1

Participants were blindfolded for the duration of this experiment in all of the following phases, immediately after informed consent was obtained. Participants did not see any of the sample stimuli until the end of the experiment during the post-experiment interview. Procedures were identical to Experiment 1 with the substitution of tactile stimuli for picture cards and instructions to look at the pictures replaced with touch/feel the stimuli. Participants were oriented to the table and where sample stimuli would be placed during tact training and sequencing tests using hand-over-hand guidance. Instead of pointing to the participant during blocking tests, the participant was instructed to place their hand on the table. A touch on the back of the hand signaled the participant needed to begin saying the alphabet.

Vocal Sequencing Tests

At the beginning of each trial, the experimenter stated, “Please attempt to put the cards in left to right order, then put your hands on your lap when you are finished, and I’m going to tell you the names of the cards in order.” The experimenter then stated the pre-arranged sequence, and then placed the tray of randomly arranged tactile stimuli to the right of the participant.

The participant then sequenced the cards on the table in front of them, and put their hands in their lap. Nonspecific feedback in the form of “thank you” or “OK” was delivered at the end of all trials. Each vocal sequencing test phase consisted of five trials, where five non-repeating random sequences of cards were tested.

Tactile Matching Sequencing Tests

During tactile matching sequencing tests, four tactile cards were arranged on the table in front of the participant within their reach, with enough space between the sample sequence and the participant to fit the participant’s tactile cards. After arranging the four tactile cards on the table, the experimenter told the participant, “Now I want you to attempt to put the cards in the order you feel, from left to right, and put your hands on your lap when you are finished.” The experimenter then placed the tray of corresponding tactile cards to the right of the participant, arranged in a random order. The sample stimuli remained on the table throughout the trial; while the participants arranged their cards, they were free to reach out and touch the sequence of sample cards. The sequences of stimuli were pre-determined and semi-random, but a sequence was not repeated within a 5-trial block. At the end of each trial, the experimenter responded with a neutral “thank you,” or “OK,” regardless of the accuracy of the participant’s performance, and began the next trial. Each tactile matching sequencing test consisted of five trials.

Echoic Training

Echoic training was conducted using the same procedure as Experiment 1.

Tact Training

Tact training was conducted using the same procedures as Experiment 1, with the addition of a blindfold for the participant and the instruction “I’ll put a card in front of you and tell you its name, I want you to feel the card and repeat the name” replaced “I’ll show you each picture and tell you its name, I want you to look at the picture and repeat the name.”

Vocal Blocking Tests

Vocal blocking was conducted using the same procedure as Experiment 1, blindfolded participants were signaled to begin saying the alphabet by a touch on their hand rather than pointing at them.

Post Experiment Interview

Participants were asked the following questions:

1. Did you use any kind of strategies to learn the name of the individual textures?

2. Did you use any kind of strategies to sequence the textures?

3. Did you talk to yourself about any of the textures?

4. Was any part of this study difficult for you?

5. Do you have anything you would like to add about your experience?

Results

Figure 5 displays the data for the percentage of accurate sequences during sequencing tests and vocal blocking tests for participants T1, T2, T3, and T4. During initial vocal sequencing tests (before they had learned the names of the stimuli), all participants responded at 0% accuracy. Performance during initial tactile matching sequencing tests was 100% across all participants. Every participant mastered echoing the names of the stimuli in one block of trials. Participants T1, T2, T3, and T4, required 11, 7, 5, and 5 trials of tact training, respectively, to master tacting the tactile stimuli.

Fig. 5.

Experiment 2 results Note. Percentage of correct sequences per each 5-trial block during vocal (closed triangles) and tactile (open squares) matching with and without vocal blocking procedures during Experiment 1.

All participants performed at 100% accuracy during tactile sequencing tests with the exception of T2 and T3 performing at 80% accuracy during the last tactile sequencing test. As for vocal sequencing tests, T1 and T3 performed with 100% accuracy, and T2 and T4 sequenced between 80% and 100% accuracy of sequences. However, blocked vocal sequencing tests showed two trends across participants. Vocal blocking performance was undifferentiated from unblocked vocal sequencing performance in participants T1 and T2. T1 accurately sequenced 100% of the trials throughout the entire experiment. Contrastingly, participants T3 and T4 scored at 20% accuracy during the first blocked vocal trials, and at 60% during the second blocked vocal trials, which were less accurate than the unblocked trials.

The participants reported in the post-experiment interview that they learned the names of each texture by either correlating the textures and names with visual features or tactile features of already familiar objects. T1 and T2 primarily relied on visual associations, whereas T3 and T4 relied on tactile associations. T1 reported, “I started to associate colors with them. Like lin was green to me for some reason. This (jom) was like a cheetah print almost.” T2 described, “This is Kevin [kev]. I just imagined some big guy playing football because it felt like a football. Jom, I forgot the name of the dude off Monsters Inc., but that’s what I thought of every time I said jom.” T4 emphasized the tactile stimulus properties by stating, “Jom is really fluffy and I love jom. This one, I feel like it feels like a nop. This is nop. This is weird now that I see them, I’m trying to identify them!”

When asked if any part of the study was difficult, participant T1 explained that memorizing each texture’s name and feeling was the most difficult part of the study, but, “after Ilearned them it wasn’t very difficult at all.” Contrastingly, T2, T3, and T4 all reported that saying the alphabet was the most difficult part of the study. T2 explained, “I just tried to echo their names in my head, but the alphabet obviously, so I just winged it.” Different levels of difficulty were reported across participants.

Discussion

The results from Experiment 2 provided a demonstration that vocal blocking procedures were only moderately effective in interfering with assumed covert verbal mediation during blocked vocal sequencing tasks, and had no effect on the performance of 1 of 4 participants (T1). These data suggest that mediational responses in tactile matching sequencing tasks may not require some form of verbal behavior in order to accurately emit the correct response, whereas vocal sequencing tasks may require verbal mediation. Vocal blocking procedures reduced responding accuracy to 20% to 60% for two participants when the sequencing task was vocal. This trend was apparent in the participants who were first exposed to tactile sequencing tasks, but not in the participants who first completed vocal sequencing tasks. These differences in performance could be a byproduct of sequence effects, but further replications with more participants are required to evaluate this conclusion.

These results suggest there may be differences in how people accomplish a sequencing task when the stimuli are tactile rather than visual, and that people may not so commonly engage in verbal mediation when “remembering” tactile stimuli. A possible limitation to this experiment is that the individuals in this study were sighted individuals and may, therefore, have more, less, or simply different strategies for remembering tactile stimuli when compared to non-sighted people. Individuals who would require tactile stimuli in an applied setting may have less robust verbal repertoires as our participants in addition to a visual impairment, so they may not be able to engage in the vocal or visual mediational strategies that may be required for correct responding (Lund & Troha, 2008). Future studies should explore how these results would translate to participants with visual, sensory, and perceptual deficits so more effective methods may be used in applied settings to maintain the effectiveness of behavior analytic services across populations. An additional limitation is presented by the two sets of tactile stimuli. We did not follow guidelines for equating the two sets of stimuli such as those suggested by Cariveau et al. (2020), and therefore the two sets could have been inherently easier or more difficult to learn to tact and/or sequence.

General Discussion

These two experiments are an attempt to extend our understanding of mediation, blocking, and private events. We found Experiment 1 to be a very clever set of procedures and have little to say about it in terms of limitations. We agree with Clough et al. (2016) that using four stimuli may be easier and therefore differentially affected by blocking mediation than with a larger array, and perhaps other researchers should arrange for more difficult tasks that might be more sensitive to vocal blocking as an independent variable. Although we attempted to match the procedures of Clough et al. (2016) in Experiment 2, the extension to tactile stimuli introduced some additional limitations. Many behavior analytic studies use arbitrary novel visual stimuli; however, we could not find any that attempted to use novel tactile stimuli. We selected fabrics with textures that were distinct from one another based on our own subjective experience feeling a variety of textures, but we did not validate these distinctions. It is difficult, perhaps impossible, to find truly “novel” textures, especially with fabric. It is possible that our participants had significant learning histories that may have generalized to any of our fabrics. For example, one could feel like a blanket that they’ve had since childhood, or one could feel like the favorite sweater of their recently separated significant other; those stimuli could evoke a great deal of private events which could help in a sequencing task. An additional limitation was in execution: failing to videotape several post-experiment interviews. Some interview responses were skillfully live-transcribed during the session by the second author, but we did not include those results as a dependent measure and could not quantify them or check those results via inter-observer agreement. Finally, we measured the dependent variable as a binary correct/incorrect rather than number of stimuli in the right place in the sequence or any other more sensitive measures of accuracy. This is a limitation, but it was a choice to make comparing our results with the study we were replicating straightforward.

Experiment 1 replicated the findings of Clough et al. (2016), and Experiment 2 shows that by changing the modality of the stimuli, the patterns of participant performance changed. Under the label of joint control, scholars in our field are showing that certain tasks, such as sequencing stimuli under the control of vocal stimuli entering into joint control with visual stimuli, require verbal mediation, which may be covert. We encourage those using these types of procedures to consider two things.

First, although experiments have shown that vocal blocking reduces performance, that is not evidence that covert mediation is required or is occurring during these tasks. If our two assumptions are that (1) verbal mediation is required for these tasks and (2) that overt vocal behavior (blocking) is incompatible with covert verbal mediation, then our results should show that the task is impossible during vocal blocking. Given four picture cards or tactile stimuli, the chance of randomly arranging them in same sequence as requested by the experimenter is 4.2%, or 1 in 24. If participants score substantially higher than that in blocked trials, we should not conclude that covert mediation is responsible for their performance in unblocked trials. The presumed covert mediation seems to aid in these tasks, but it is not a requirement for this performance. Alternatively, we could reject the second assumption and instead conclude that saying the alphabet may not be incompatible with covert mediation. Gutierrez (2006) said the participant whose performance was less affected by blocking was singing the blocking vocal very slowly, “possibly allow[ing] her to rehearse the sequences covertly” (p. 189) and similarly Harman et al. (2021) speculated that “participants could alternate between mediating verbal behavior and the vocal alphabet” (p. 73). These call into question the common practice of using vocal blocking to study mediation.

Second, as far as we can find, extant joint control experiments have included visual and auditory stimuli. One could argue that hand-sign mediators (Clough et al., 2016, Experiment 2; Lowenkron, 1988) have a proprioceptive component, but another explanation is that the visual stimulus of one’s hand is controlling the responses. In Experiment 2, we changed the modality of the stimuli, but we should not predict a change in performance, since there is nothing in the joint control account that says the stimuli have to be seen vs touched, tasted, etc. However, we did observe a change in performance, and that should give us pause to re-evaluate our assumptions and widen our methodologies to examine different types of tasks when seeking to learn more about covert behavior. The patterns observed in the results of joint control studies using visual and auditory stimuli could be due to common learning histories around visual tacts rather than characteristics of joint control itself. We commonly tact the things we see, and we commonly learn to put things in order that we see and tact. Far less common, perhaps, is putting things in order by their texture, or learning tacts for texture in the first place. Visual tacts are far more common than other sense modalities. Take, for example, the word “lemon” and the private events it evokes. You may imagine an image of a lemon and/or the sour taste of a lemon, but may be far less likely to experience a private event of the texture of the skin of a lemon. Think of the relative proportion of tacts for visual stimuli compared to tacts of textual stimuli in your repertoire. Although many individuals have a repertoire of texture tacts (e.g., rough, smooth, bumpy, sticky), in comparison to tacts of objects (physical things in one’s environment), it’s difficult to come up with many tacts for which the most relevant stimulus is its texture (exceptions: sandpaper, glass). Participants in this study (and people in general) may then have less developed, less sophisticated, or just qualitatively different problem-solving strategies for dealing with tactile stimuli, and mediation is less of a factor as a result.

We would not argue that joint control does not occur as it has been described. There are times when we rehearse a list, series, or instructions, and then respond under the control of a discriminable jump in response strength when we see or create stimuli that correspond with our rehearsal in some way. Nevertheless, just because that is a phenomenon that occurs does not mean it explains a larger portion of complex human behavior. The understanding of private events, especially covert verbal behavior, is an important frontier of behavior analysis. Rather than settling on one tool or paradigm to study covert verbal behavior, we should be developing, testing, and critiquing a variety of methodologies to ensure we continue to have the best understanding of the behavior of organisms.

Funding

This study was conducted with financial support from the Kirbo Scholar Grant at Berry College, awarded to the second author.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

The Institutional Review Board for Human Subjects Research at Berry College examined the procedures of this study and found them to be in compliance with all standards and regulations. All participants gave informed consent to participate in the study.

Conflict of interest

We have no known conflict of interest to disclose.