The study of language phenomena in metacontingencies is primarily based on Skinner’s (1953) analysis of verbal behavior and Glenn’s (1991) analysis of cultural practices. From these vantage points, verbal interactions are understood in terms of their consequences on the environment, such as how the speaker’s behavior or verbal stimuli affect recurrence of behavior products. It is also assumed that language phenomena support cultural selection in the forms of effective modes of communication via tacts, mands, and autoclitics (Skinner 1957, 1986). Moreover, the communicative function of language helps propagate ways of living and shared modes of behaving useful for cultural survival in established civilizations. The selection of verbal activities is said to be closely associated with survival practices, including the support of a myriad of cultural practices (Glenn 1991). In sum, the verbal behavior of individuals affects modes of communication as well as transmission of practices that are useful for cultural survival. The role of language in metacontingencies, then, is closely associated with propagation of effective IBCs generating viable APs (Glenn et al., 2016).

Five metacontingency studies have examined the role of language in selection of culturants by measuring frequency, duration of classes of verbal utterances, and generation of APs by group members (Costa et al., 2012; Hosoya & Tourinho 2016; Sampaio et al., 2013; Smith et al., 2011; Soares et al., 2018). For the most part, these researchers have examined verbal behavior in metacontingencies under the assumption that its primary function is to support transmission of practices among individuals or as consequence to establish operant behavior associated with generation of accurate APs. Hosoya and Tourinho (2016) examined the punitive and reinforcing functions of verbal utterances in selection of IBCs generating correct APs in two microcultures. Their study used an ABAC design in which type of contingency (operant versus meta) and either allowing or restricting verbal interactions among participants were manipulated. Also, these verbal interactions were limited to the first five minutes following change of generation (i.e., new participants replacing old ones). Hosoya and Tourinho’s findings indicated that verbal interactions may function as consequences for selection of IBCs. It is important to note that the extent to which verbal interactions were needed in the development of IBCs was not addressed in this study (Hosoya & Tourinho 2016).

Sampaio et al. (2013) analyzed differences in selection of IBCs generating APs in triads that were allowed to communicate versus those in which communication was restricted. Results from this study showed covariation in quantity of utterances and IBCs, suggesting that the effects of consequences on IBCs are mediated by verbal utterances (Sampaio et al., 2013). It was not possible to make further conclusions about the participatory role of language in selection of IBCs due to a small sample (four dyads total, one in the communication group). Verbal interactions of triads were only reported anecdotally but no functional analysis of language was provided. However, Sampaio et al. noted that some participants developed leader roles during the task, which bespeaks of the importance of accounting for rule-following and individuals’ histories of verbal relations to explain variations of IBs inside of IBCs. More recently, Soares et al. (2018) examined effects of verbal and nonverbal consequences (i.e., different messages appearing in each participant computer screen) over production of culturants (i.e., IBCs and APs). Their results suggest that verbal messages have a selective effect on production and recurrence of IBCs.

Overall, the participation of language in metacontingency has been studied by measuring quantity of speaker utterances, analyzing transmission of instructions among participants and consistent generation of APs, and examining the effects of different forms of rules on participants’ performance. These measurements of language phenomena have been developed primarily from Skinner’s (1957) definition of verbal behavior, which applies to the speaker and leaves the listener as a secondary element whose participation may only be limited to mechanical action. However, Skinner also acknowledged that we need to account for the total verbal episode in order to explain the “verbal” aspect in the interaction (1957, p. 34). In short, verbal interactions may be understood as the functions of the behavior of speaker and listener and the verbal properties of the setting (e.g., instructions, symbols, schematics, etc.) in which these relations occur.

For example, communication inside organizations may occur in the form of verbal products (e.g., rules passed from managers to employees) and depends on a history of verbal relations with those products (Houmanfar et al., 2009). Individuals may act as listeners in different ways depending on how rules affect their relational responding, such that some individuals may form rules allowing their behavior to be more susceptible to changing contingencies, while others may behave under the basis of a history of socially mediated consequences (Ghezzi et al., 2020; Rafacz et al., 2019)

Communication also occurs in the form of referential interactions, wherein speaker, listener, and referents participate in the context of a cultural environment. Referential interactions refer to a person’s simultaneous reactions to a listener and a referent—the thing being talked about—, under a specific setting (Kantor 1977). The listener may be another person or the speaker herself (talking to oneself). A referent may be present or absent, concrete or abstract, existent or nonexistent. Setting conditions may be social, physical, psychological, or a combination of the three. These conditions affect the stimulus functions and response functions in an interaction. An analysis of verbal interactions can assist with our analysis of response variabilities in IBCs due to individuals’ similar or shared histories of reinforcement.

The secondary adjustment function of verbal behavior (Kantor 1977) which captures the shared histories of individuals, presents another important feature of the dynamic interaction of individuals inside of IBCs. To define this additional function of verbal behavior, an emphasis is given to the use of speech to achieve determined results beyond referential adaptations (i.e., completing a task). In this context, verbal behavior is analyzed in terms of ways by which (e.g., persuasion, humor etc.) the speaker evokes some action on the part of the listener, (e.g., cooperation). In other words, the analysis of verbal interactions constitutes not only the orientation to the task, but also individuals’ adjustments to other factors in the cultural environment (seeHoumanfar et al., 2010; Houmanfar et al., in press).

Measurement of secondary verbal adjustments may allow researchers to examine the relationship between social interactions and the contextual factors (see discussion of cultural milieu by Houmanfar et al., 2010 & in press) under which they occur. In the context of individuals interacting inside of IBCs, the speaker may indicate current, past, or future iterations of the task (i.e., lineages) and refer to an event that has occurred or is imaginary. The majority of metacontingency research thus far may be said to be derived from a similar experimental arrangement (i.e., Wiggins 1969; seeZilio 2019 for a complete review of experimental metacontingency literature) in which participants interact in a programmed sequence of steps to generate conjoint or aggregate products that must meet some selection criteria. In this process, the particulars of objects, events, or persons may include: (a) elements of or composite aggregate product, (b) group members, (c) participants’ roles (if any), (d) other people (external to the group), and (e) institutional (if speaker’s referent is about activities related to consumer demands, instructions, or rules inherent to the task). A majority of these factors may play a role in interpersonal and interprofessional dynamics, that is, in the recurrence (or nonrecurrence) of IBs that in turn affect the recurrence of IBCs and associated APs. In short, how verbal behavior participates in metacontingencies, in terms of the communication and secondary adjustment function of verbal behavior, warrants empirical examination.

Smith et al. (2011) examined effects of different forms of rules (explicit, implicit, and no rule; see the taxonomy of rules developed by Peláez & Moreno 1998) presented to dyads throughout their performance during an organizational task, conceptualized as a five-term metacontingency analog. Individuals in dyads generated APs consisting of a rectangle and a circle by choosing correct color and size for each shape. Dyads achieved correct APs by correctly choosing, in a sequential manner, first the color and size for the rectangle, and later the same dimensions for the circle. The dimensions of the shapes were specified (in different degrees of explicitness) in the rule provided at the beginning of each trial. Smith and colleagues found that interindividual performance (measured as accuracy and number of APs) is influenced by ambiguity of the rules presented to participants. Their results indicate that higher percentage of correct APs are observed when working under explicit rules and higher response variability occurs under implicit rules. The authors also reported that dyads developed patterns of verbal interactions on the basis of their shared history by the end of the experimental task, however, the types of patterns were not reported in this study.

Smith and colleagues’ (2011) research indicated two important aspects regarding the development of IBs inside of IBCs: (a) verbal interactions between individuals are needed to perpetuate occurrences of IBs and variations thereof; (b) generating correct APs does not necessarily indicate selection of associated IBs (Smith et al., 2011). We extended these findings by examining ways the level of ambiguity of antecedent factors(three degrees of explicitness in rules were presented to participants throughout the experimental task), may affect recurrence of IBs in dyads. We addressed two limitations pointed out by Smith et al. (2010) about their procedure. First, the time-criterion used to change conditions was replaced by a stability criterion of APs production such that variability in pace and cycle completions per dyad were unaffected by the procedure. Second, communication among participants in dyads was experimentally controlled, which permitted between-group comparisons in terms of the effects of communicating on interindividual performance. In addition, we addressed a third limitation in Smith et al.’s study concerning the discriminative feature in their task—individuals were able to observe their partner’s choice. In this experiment, the computer screens only displayed the individual’s own choices (color and size), and not their partner’s choice. This allowed the systematic account of presence or absence of verbal interactions in IBCs. In short, this study sought to (a) examine effects of increasingly ambiguous forms of rules presented to participants in their performance, and (b) investigate the participation of verbal behavior in acquisition and maintenance of IBs. We extended Smith et al.’s findings by examining effects of three degrees of explicitness of rules presented to participants in their performance and verbal interactions. The verbal behaviors of participants were analyzed as total interactions, wherein speaker, listener, and functions of stimulus objects or events constituted the main elements to analysis referential interactions in designed IBCs.

The independent variables included the following rules that were presented to the participants at the beginning of each trial: (A) high explicit–HE; (B) medium explicit–ME; (C) low explicit–LE. The rules specified how participants should respond in order to correctly complete APs on which they worked (see Appendix B). High explicit (HE) rules specified precisely how participants should respond in terms of size and color of forms (e.g., “Your product must be PURPLE in color and within Q size range”); medium explicit (ME) rules only specified color whereas size was vague (e.g., “Your product must be ORANGE in color and SOMEWHAT SMALL in size”); and in low explicit (LE) rules specifications of both color and size of forms were ambiguous, and fairly open to interpretation by participants (e.g., “Your product must be the color of FLAMINGOS and within LOW size range”). Following Peláez and Moreno’s (1998) rule taxonomy, rules varied in degrees of explicitness such that the more implicit the more elements were implied (the discriminative stimuli, context, or consequence). The rules in all conditions described consequences in a similar way but differed with respect to the clarity of which they describe stimuli relevant to the orientation required for a particular response. HE rules contained explicit/explicit components (explicit color and size), ME rules contained explicit/implicit components (explicit color and implicit size), and LE rules comprised implicit/implicit components (implicit color and size). See Table 1 for a complete definition of each rule type.

Table 1
Operational definitions of different levels of explicitness of rules

The dependent variables included frequency of correct responses and APs across experimental conditions and time to complete three consecutive APs for the same rule. The dependent variables associated with verbal interactions included the frequency and types of secondary verbal adjustments such as persuasive statements and humor during five-minute intervals (seeTable 2) across experimental conditions. The method to analyze verbal interactions was based on Bijou et al.’s (1988) coding system of referential interactions (seeSmith et al., 2011; Smith et al., 2012; Johnson et al., 2010).

Referential interactions

Referential interactions consist in a speaker referring something, someone, or some place, to a listener who acts in some way to the referred circumstances.

Secondary verbal adjustments were identified when the speaker’s verbal actions had an additional effect on the listener’s behavior, such as to persuade or use humor to evoke listener’s cooperation. Each type of secondary verbal adjustment was defined in terms of its observed effect on the listener (seeTable 2).

Table 2
Operational definitions of elements in referential interactions

Interobserver agreement (IOA) was determined by point-bypoint agreement between a principal rater and a second rater. Both raters independently analyzed data pertaining to the verbal interactions. IOA was obtained for 30% of the total number of verbal interactions per dyad with an agreement ranging between 86 and 98% for all dyads analyzed from the verbal groups.

Experimental design

A counterbalanced reversal design was employed for four groups: Nonverbal (NV): HE-LE; Nonverbal (NV): LE-HE; Verbal (V): HE-LE; and Verbal (V): LE-HE. A total of 27 dyads were randomly assigned to one of the four experimental groups[1] (seeTable 3). The primary function of the reversal design was to compare the differential effects of the three forms of rules (HE, ME, & LE) on development of IBs and production of APs. Additionally, the counterbalanced sequences were employed to assess any potential sequence effects.

Table 3
Dyad assignment in four experimental groups

Note. HE = High explicit; ME = Medium explicit; LE = Low explicit; V = Verbal; and NV = Nonverbal.

Once participants signed consent forms, instructions were handed to them and also read by a research assistant. The instructions advised participants that they will be working together as a team to design and generate products with two specific dimensions and that they would each earn 10 cents for every three consecutive correct product they produced (see Appendix A). They were told that their job was to find the best strategy to understand the rules presented throughout the study in order to generate revenue. The researcher noted that their revenue would be tracked on each of their screens and they would receive their earnings at the end of the experiment. Once the task ended, participants had to complete a questionnaire addressing questions about the task.

As in Smith et al. (2011), participants took turns responding in each trial to generate a product comprised of a circle and a rectangle with specific dimensions. During a given experimental trial participant one (P1) had to first choose the color of the rectangle (using a mouse click), after which participant two (P2) selected the size for the same part; it was then P1 who had to choose the size for the circle, followed by P2 choosing the color for the circle part (seeFigure 1). The four-response sequence constituted one trial and no one of the responses could be emitted until the prior response had been completed. Upon completing a trial, participants received a message specifying whether their product was correct or not. If the product was not correct, another message was provided to each participant on their computer screens saying: “Sorry, either your product or your partner’s product was not within specifications”, and the trial restarted with the presentation of the same rule; if product was correct then the following message would appear on their screens “Good job. You and your partner created the product within specifications. Make more products to earn money.” Based on the pilot data, a criterion of three consecutive correct APs per rule or 20 trial attempts were required to move from one rule to the next within a condition. An experimental condition ended after the completion of 9 APs or 60 trial attempts. The task terminated after 2 hours or after dyads completed 18 rules (3 per condition). All dyads included in this analysis were exposed to the same number of rules (i.e., 18 rules, see Appendix B). Dyads that reached the 120-minute mark but did not complete all the rules were payed the amount of money earned and dismissed.

All colors and sizes of rules were presented in a randomized order phase to phase such that no one color, or size, could be part in consecutive rules. Selection of rules for each condition was the same across groups such that all dyads were exposed to the same rules in same order, according to the experimental group. Rule criteria (color and size) were the same for both participants in each phase. Participants could choose from 18 color options, three variations (hues) per color. Participants selected the color for their product by clicking over one of the options and clicking submit. Different from selecting color, the size was selected by moving a scroll bar such that a portion (i.e., size) of the part was filled varying along a continuum. The only difference between phases were the rules presented at the beginning of each trial and a change in consumer demand (correct size and color). Phase changes were therefore signaled by the presentation of a new form of rule. After completing a questionnaire, participants were shown a screen telling them to inform the research assistant that the task had ended.

Figure 1
Sequence of events in a given trial

Note. Sequence of events in a given trial. The four responses, two per participant, are shown in a sequential order of completion. Colors were selected from a discrete range of 18 options, while size was selected by filling an interval using scrolling option.

The concept of the cultural milieu may offer a point of entry to examine how cultural environmental factors (psychological, anthropological, ecological/biological) affect the acquisition of IBs in the context of IBCs (Houmanfar et al., 2010). According to Houmanfar et al.’s elaborated account of the metacontingency (2010, in press) the selecting environment constitutes the cultural milieu which includes the consumers of the aggregate product in addition to overarching cultural beliefs, material resources, governmental policies, rules, traditions, morals, institutions, technological progress, and environmental competition. More specifically, the cultural milieu constitutes the collection of the distinctive stimulus functions such as post covid-19 pandemic changes in training procedures, policies associated with social distancing, wearing mask, and stated political values during the coinciding presidential election in the United States of America (USA). These stimuli influenced the acquisition and maintenance of IBCs such as Applied Behavior Analytic (ABA) training processes in the United States (USA), associated IBs (employees’ level of stress and implicit political biases affecting interprofessional interactions), and behaviors of individuals (students who received ABA training in USA) that interact with the associated aggregate products (e.g., training packages, workshops, etc.). Behaviors of consumers in this context can be discussed as macrobehavior (socially learned operant behavior observed in the repertoires of several members of a cultural system) of knowledgeable ABA practitioners who work in school districts and the associated cumulative effect demonstrated by improved cases of maladaptive behaviors of children in selected school settings (i.e., macrocontingency).

Future metacontingency studies may examine the effects of different cultural milieu factors on the dynamics of IBs in IBCs, and the impact on their generation of aggregate products. For example, experiments have shown that under situations of choice between working in shared contingencies or alone, the content of instructions delivered to participants prior the beginning of experimental session affected their selection of contingencies (Pacheco-Lechón & Carpio 2014). According to these experiments, individuals worked together despite changes in external selection processes (e.g., losing points or money) in conditions that provided shared contingencies (Pulido-Avalos et al., 2015, Rangel et al., 2015).

Additionally, based on the data provided herein, we may also suggest that the standard sequential task in metacontingencies mainly promotes patterns of verbal behavior established with respect to achievement criteria (e.g., cooperation, persuasion). The “interlocked” element of IBCs may also be manipulated by altering the sequential nature of the experimental task and, in doing so, verbal problem solving, and factors associated with the shared contingencies can be determined.

The major limitations of this study were the number of dyads and their unequal distribution across experimental conditions. In addition, although the experimental task employed allowed the manipulation of IBCs and measurement of IBs, a more complex task may better approximate socio-cultural situations as they occur in natural settings, including how different types of verbal phenomena (e.g., organizational rules, verbal interactions) participate in the development of certain kinds of IBs in IBCs. Future studies should also consider more precise ways to assess the level of ambiguity of the rules (low, medium, and high). It is possible that the difference in number of incorrect products observed across groups may have been influenced by the perceived ambiguity of the rules, including the order in which they were presented (see Appendix C). The results indicating incorrect products increased as ambiguity decreased in some conditions for some dyads, suggests that some rules may have been perceived as more ambiguous than we anticipated. This issue may be prevented by conducting a preassessment of rules or talk aloud procedures to determine their level of ambiguity.

As discussed by Houmanfar et al. (in press), at the psychological level, the term IB highlights the critical role that individual participants’ histories play in the interaction of individuals within a given IBC and, ultimately in the selection process associated with the metacontingency. Said another way, an analysis of the close interrelationship between psychological and sociological factors in metacontingencies may lead to a better understanding of social interactions, including how structure (e.g., IBCs) interacts with networks of individuals (e.g., IBs). For example, Molm et al. (2009) demonstrated that some structures of exchange relations (i.e., reciprocal) and contextual factors (perceived risk and uncertainty) influence the development of certain psychological factors (e.g., resilient trust) in interpersonal relations. The metacontingency has the great potential for providing the platform for analysis of these types of interlocked factors in complex behavioral systems.