Skinner’s initial career was marked by the objective of establishing a science of behavior based on the natural sciences. This objective implied a rejection of metaphysics[1] and a commitment to the ideas that are characteristic of Newton’s classical mechanical model such as determinism and the necessity relation between events (e.g., Cruz & Cillo, 2008; Laurenti 2009; Micheletto, 1995; Moxley, 1999; Sério, 1990). These ideas became evident with Skinner’s proposal for explaining behavior using the reflex concept and marked the elaboration period of his explanatory system during the 1930s.

The concept of reflex in the description of behavior had a crucial role, because it reflects the kind of relation of dependency between events employed in Skinner’s nascent system: “The reflex is important in the description of behavior because it is by definition a statement of the necessity of this relationship” (Skinner, 1931/1999, p. 433). Hence, there is a necessary and unequivocal relation between stimulus and response. It is interesting to notice that at this point Skinner already adopted Mach’s idea that explanation is the description of functional relations, the same idea that later on will lead to an alternative concept of the dependency relation between events that is characteristic of a selectionist perspective. At this early stage, however, the replacement of the notion of cause by that of function did not seem to contradict the expressed necessity of the reflex relation.

Experimentally demonstrating the necessity that characterizes reflex relations encountered certain restrictions that were intrinsically related to the issue of the variability consistently found in the course of successive elicitations. Without jumping ahead to the analysis of the initial treatment that Skinner’s system gave to variability, it is noteworthy that, according to Skinner (1931/1999), these limitations that apparently could put the definition of reflex at risk in terms of stimulus-response necessity were interpreted as methodological limitations, because it was impossible to identify all the variables that would be related to the behavior studied.

Faced with difficulties imposed by the variability observed in reflex relations, it was precisely in the attempt to assure the regularity in terms of necessity between events that a second research area was proposed to describe the changes that occur in several aspects of the correlation on the basis of one more law. Therefore, the formulation of a second type of law, entitled secondary, had the function of guaranteeing the coherence of the Skinnerian explanatory system, since it implied supporting an explanation for behavior in terms of an inexorable relation between stimulus and response. Notions such as conditioning, emotion, and drive were present as changes in the strength of the reflex, reflecting orderliness in the variation itself and thus ensuring the adequacy, from a scientific standpoint, of describing behavior through the reflex principle.

The concern with identifying regularities in changes in reflex strength became central to Skinner’s system and it was in the context of describing conditioning as one of the secondary laws that the notion of the reinforcing stimulus arose, a notion which, as Sério (1990) argued, began to be regarded as a defining property of the conditioning process. The first step in the formulation of the selective role of consequences seems to have been taken during the debate over the two types of reflex, in 1935, when Skinner observed that “the conditioned response of Type I does not prepare for the reinforcing stimulus, it produces it” (Skinner, 1935/1999b, p. 463). Hence, already in this statement, Skinner recognized that there is a type of behavior that operates in the environment and that produces reinforcement [2]. Although the quotation still does not make explicit the action of a reinforcing stimulus on the probability of future responses, it was suggested in other excerpts that the role of conditioning, through reinforcement, is to alter the strength of reflex relations: “In Type I stimuli may be divides into two classes, positively and negatively conditioning, according to whether they produce an increase or decrease when used as reinforcement” (Skinner, 1935/1999b, p. 459)

The emergence of the term ‘operant’ (Skinner, 1937/1999) is emblematic of the development of a selectionist perspective, because the Skinnerian proposal became more flexible from that moment onwards. This flexibility is achieved by identifying another type of behavior, one which demands other principles or laws, different from those that were compatible to the concept of reflex. As first articulated (Skinner, 1937/1999), the reinforcing stimulus was not yet exclusively related to a consequent event, and there was no mention of a selective effect nor of the term ‘consequence’. Nonetheless, it was the first time, at least as far as the analyzed texts are concerned, that Skinner referred to the procedure of originating new responses from undifferentiated responses through successive approximation to a specific programmed response. In Skinner’s (1937/1999) words: “The response in its final form may be obtained by basing the reinforcement upon the following steps in succession. . .” (p. 469).

From this quote it can be argued that there is at that point in the development of Skinner’s system the idea, at least implicitly, of a selective effect of the consequent event, the reinforcing stimulus, since it already has been suggested that a new response can be established via reinforcement, a fact that shows selection in the process. However, as noted by Serio et al. (2000), there are no direct references to the selective role of consequences produced by the operant response and, therefore, that function was not yet formulated. Consistent with these observations, Peterson (2004) stated that Skinner’s reference in 1937 to the development of a final response through reinforcement was speculative. In other words, it is probable that such an idea in this period was still a hypothesis, as Skinner (1979) seemed to suggest in his autobiography: “I do not remember actually shaping lever-pressing in such explicit stages, but I was sure it could be done, and I had certainly changed the ‘value of a single property’ through successive approximation in producing very forceful responses” (p. 185).

On the publication of his first book (Skinner, 1938), by retaking the operant notion Skinner reiterated the idea that a response may occur without any specific observed antecedent event and still be described scientifically. This treatment of operant behavior shows an initial attempt to insert in the scientific field a type of relation that would be more compatible with the explanatory selectionist principle proposed years later. However, the defense of the adequacy of employing the term “reflex” to describe both types of behavior and, at the same time, the acknowledgement of the distinct nature of operant behavior shows inconsistencies, which are noted in the following statement:

Operants, as predictable entities, are naturally isolated last of all because they are not controlled through stimuli and are subject to many operations. They are not obviously lawful. But with a rigorous control of all relevant operations the kind of necessity that naturally characterizes simple reflexes is seen to apply to behavior generally. (Skinner, 1938, pp. 25-26)

Likewise, Skinner (1938), while commenting the notion of reflex, continued to emphasize that: “Somewhat more generally, the term applies to a way of predicting behavior or to a predictable unit. In this broad sense the concept of the reflex is useful and applicable wherever predictability may be achieved” (p. 439).

These quotes suggest several further points. First, at the same time that Skinner (1938) referred to reflex as a necessary relation between events, he made the definition more flexible, assuming that the term is applicable, ultimately, to any behavioral unit that is susceptible to prediction and control. This suggests that the operant already was recognized a type of behavioral relation that cannot be described in terms of necessity, but that, in spite of it, exhibits regularity. Nonetheless, even emphasizing the specificities of the operant and admitting at times that the term ‘reflex’ in its original meaning is applicable to the respondent only, Skinner also pointed out that, with a certain level of control, the type of relation imposed by the reflex principle is a characteristic trait of any behavioral relation, respondent or operant.

Sustaining this type of dependency relation between events, which, however, contrasts to a selectionist perspective, proved to be harder and harder with the explanation of specific characteristics of operant behavior. For example, Skinner (1938) explained that the antecedent stimulus, in this case, no longer inexorably ensures the occurrence of the response, as a necessity relation between events presupposes: “In Type R the reinforcement is contingent upon the occurrence of the response, and there is no certain way of obtaining a response at a given time during the presentation of a discriminative stimulus” (p. 270). Thus, Skinner, while attempting to apply the reflex notion for explaining behavior in general, presented, with the concept of operant, evidence of a transition to the adoption of an alternative type of dependency relation between events, more compatible with selectionist ideas.

The debate over the concept of reflex surrounded by the notion of classes of events, retained in the 1938 book, brings out the idea of the selective role of the reinforcement stimulus again. In admitting variations in the level of response class, Skinner (1938) noted that, in the context of operant behavior, a response class is defined by the reinforcement conditions. This fact, then, suggests that reinforcement has the role of selecting specific classes of responses, in the sense of increasing their probabilities of occurrence and decreasing the probability of responses belonging to other classes. If this interpretation is correct, it is possible to conclude that the first characteristic of the selectionist principle is almost explicit at this point in Skinner’s work.

The selective role of consequences becomes even more evident when Skinner (1938) discussed the differentiation of a response. He made it clear that because reinforcement allows variability within the class, that is, between the parameters of the responses that belong to the class, it becomes possible to differentially reinforce a certain property of the response, such as its intensity or duration:

Although a relatively constant mean slope is maintained during reinforced responding, there is some variation in the force of single responses, and it is therefore possible to reinforce differentially with respect to this property. . . . When responses are differentially reinforced with respect to their intensity, the relative frequency of strong responses immediately increases. (p. 314)

Therefore, it is noted that there is a process in which reinforcement acts on within-class variations, resulting in the maintenance of the strength of responses that presented properties specified by the reinforcement. Although the term selection was not used and there was no explicit reference to the selective role of the reinforcing stimulus, the idea of selection via reinforcement of certain behavioral variations and not others also appears as already implicit in the context of this discussion. Furthermore, Skinner (1938) emphasized the role of reinforcement in strengthening behavior. On this point, Skinner (1979) reiterated that: “reinforcement emphasized strengthening” (p. 97). So this means that the presentation of certain environmental stimuli increases the rate of responses that are similar to those reinforced, while other are eliminated, a fact that can be seen as evidence of the selection process.

In the context of a debate concerning some properties or aspects of behavior to affirm possible contributions of the experimental analysis of behavior to neurology, Skinner (1938) referred to the retroactive action of the reinforcing stimulus: “Schemes for explaining Type S in terms of simultaneously active paths are inadequate for Type R, which presents a special problem in the apparently retroactive action of the reinforcement” (p. 431, emphasis added). Thus, it is possible to say that with a Type R reflex, reinforcement functions to affect behavior retroactively, increasing the probability of future occurrences. The use of term ‘apparently,’ however, may suggest that this recognition was still somewhat incipient.

The end of the 1930s was marked by a change in emphasis from respondent to operant, with the prominence of what the organism does in the environment, and controversy over the applicability of the term “reflex” to both types of behavior. In the 1940s, Skinner fostered debates more focused on human behavior, contributing even more to the formulation of the selection process. In 1945, for example, in discussing the operational analysis of psychological terms, Skinner began to point out the reinforcing function of the verbal community: “The individual acquires language from society, but the reinforcing action of the verbal community continues to play an important role in maintaining the specific relations between responses and stimuli which are essential to the proper functioning of verbal behavior” (Skinner, 1945, p. 272).

Furthermore, the role of reinforcement history for explaining behavior also becomes increasingly evident. Skinner (1947/1999) started referring to behavioral history as a fundamental determinant, along with the individual’s genetic constitution, of human behavior. This kind of reference brings out the historical aspect of an explanatory principle based on a selectionist perspective. This is because the latter perspective emphasizes the selective function of past consequences of behavior. Therefore, Skinnerian selectionism suggests that we look at the reinforcement history of the individual behavior, because, for this explanatory mode, the immediate events occurring when the response occurs are not sufficient to explain behavior.

An alternative notion of a dependency relation between events, more compatible with the selectionist principle also begins to appear, as when Skinner (1947/1999) used probability as a key word in his science: “The end term in a theory of behavior, in short, is the probability of action” (p. 324). Probability is a fundamental concept for the formulation of the selective role of consequences, given that this role is described in terms of its function in altering probabilities of future occurrences of operant responses. Thus, from this perspective, control via reinforcement can be interpreted as always being probabilistic.

In the 1930s, the notion of probability was absent, at least in part because it was incompatible with the emphasis on the necessity of the reflex relation. Therefore, the idea that reinforcement alters reflex strength and, consequently, its rate of occurrence, supplied indications of a selection process, but made no reference to the probabilistic nature of the action of this process. In the 1940s, even though he had pointed out the concept of probability, Skinner (1947/1999) continued to defend the position that human behavior is completely determined, appealing to determinism as scientific premise. The term ‘determinism’ encompasses a multiplicity of definitions and different levels of analysis. Moreover, depending on the definition adopted, a specific connotation to the concept of probability is given (see Laurenti, 2008). Skinner (1947/1999) embraced determinism as a declared commitment and treated unpredictable aspects as products of the impossibility of identifying all the variables determining behavior. Consequently, the concept of probability was, at that time, used as a methodological tool to handle this constraint on the scope of operant behavior, and not as a natural aspect of behavioral relations, as would be assumed from a selectionist stance (see Leão & Carvalho Neto, 2016).

The emphasis on prediction and control marks the end of Skinner’s works during the 1940s. With the publication of Walden II, the function of the reinforcing stimulus in altering probabilities was clearly described. It was also in this book, that Skinner (1948) used the verb ‘shaping’ (p. 105) for the first time, while discussing techniques and practices of behavioral engineering. The latter involved shaping the behavior of members of a group in order to benefit the well-being of all. It is curious that in his autobiography, Skinner (1979) wrote about an event from 1943, during a project carried out in the war period, with Norman Guttman, in which Skinner, together with Keller Breeland, on a day off, shaped the behavior of a pigeon to move a wooden ball with its beak as far as its cage. Regarding this accomplishment, Skinner declared: “I remember that day as one of great illumination. We had discovered how much easier it was to shape behavior by hand than by changing a mechanical device” (p. 268).

The remarkable effect of the use of the successive approximation technique mentioned in relation to the above episode shows that, as Peterson (2004) argued, this was possibly the first time that Skinner shaped behavior through observation and operated the reinforcer- release device with his own hands, without resorting to a mechanical device. Thus, this allows us to suggest, firstly, that while writing Walden II, Skinner already was thinking of shaping as a specific process of behavioral change, and, secondly, perhaps because of this, the term “shaping” replaced the terms used in the 1930s, that is, ‘response differentiation’ and ‘successive approximations’. Furthermore, the presence of the notion of shaping reiterates the argument presented here that, although it had not been overtly mentioned, a selective process evinced by operant behavior already had been implicitly suggested.

Another important point that corroborates the implicit idea of a selective role of consequences is the argument that Skinner (1948) presented concerning the conflict between the good of society and the immediate good of the individual. Without focusing on cultural planning to benefit the group, aspects of which shall be discussed in more depth as the last topic below, it is worth noting the present idea of immediate consequences, reinforcing for the individual, but also of delayed consequences, which are important for the survival of culture. Therefore, despite the term ‘consequence’ having been used only colloquially at this point in Skinner’s work, a preliminary idea of selection as the mechanism for the evolution of behavior, later systematized in different dimensions, is implicit in this debate on ensuring that the individual is also under the control of the group’s well-being.

In analyzing evidence for the first appearance of the selectionist mode in the first decades of Skinner’s system, his difficulty in sustaining a mechanistic position, such as describing, without metaphysical inferences (Micheletto, 1997), necessary relations between events, was noted above. During these first decades, the transformations that influenced the physical sciences at the end of the nineteenth century directly impacted Skinner’s treatment of the variable nature of behavioral phenomena. In his first article (Skinner, 1930), one can see how variability represented, at the time, a threat to the goal of scientifically explaining behavior through the reflex concept (see Leão, Laurenti, & Haydu, 2016).

Variability traditionally had been treated, from a mentalist perspective, as a byproduct of mental entities, such as the creative mind and personality, as well as being offered as evidence of how incontrollable and unpredictable psychological phenomena are (Sério, Andery, & Micheletto, 2005). Thus, eliminating the reflex concept in terms of the necessity between events as the basis of behavior invited the invocation of metaphysical constructs to explain such variability. So when Skinner proposed explaining behavior scientifically, the recognition of its variable nature could suggest an approximation to metaphysics and pose a risk to his scientific program. This context suggests that Skinner’s main concern in this first decade was to demonstrate that the variability in reflex relations would not pose a risk to the regularity of behavior so long as the variation itself was describable scientifically. He said that:

The resulting variability of behavior is typical of the sort which has led to protestations of the inadequacy of the reflex concept. But variability in the observed as against the predicted does not question the validity of a law if the variability is itself lawful. (Skinner, 1930, p. 434)

While observing that in a reflex relation a stimulus does not always inexorably elicit a response, Skinner (1930) started to evaluate what conditions might be responsible for the change in the strength of such stimuli. Consequently, variability was interpreted as a byproduct of certain facilitating conditions of elicitation that alter the strength of the reflex relation. In fact, as Palmer and Donahoe (1992) argued, the methods employed by Skinner in this early work had been focused on the relation between these conditions and the resulting change in the reflex strength, in order to explain the variability in behavior. Nevertheless, this treatment reflects an attempt to eliminate variability, by describing it in terms of another law, rather than recognizing it as a natural characteristic of behavioral relations. According to Skinner (1930):

In this instance, for example, it should be possible, once having determined a measure of the ‘strength’ of these reflexes, to investigate the conditions under which this strength changes and to eliminate [emphasis added] the variability by describing it in a further law. (p. 434)

This type of treatment for variability was what led to, as mentioned in the last section, the elaboration of a second field of investigation centered on secondary laws that describe changes in reflex strength as a function of other variables and which would explain the apparent variable aspect of the reflex relation: “In the behavior of intact organisms the apparent variability of specific stimulus-response relationships emphasizes the importance of laws of the second sort” (Skinner, 1931/1999, p. 438). From this, Skinner’s system assured the demonstration of behavior determination through the reflex principle, free from inference from any event of alternative nature, and therefore, not physical, at the core of his explanatory proposal.

With the emergence of the concept of the operant, there was an acknowledgment of a certain spontaneity in any behavioral relation, in light of the absence of an eliciting stimulus at the moment of occurrence of this type of behavior. As Skinner (1938) noted, the “dynamic laws of operants enables us to consider behavior which does not invariably occur under a given set of circumstances as, nevertheless, reflex (i.e., as lawful)” (p. 25). However, Skinner’s explanation of the spontaneous aspect appeals to the fact that operant behavior involves an unlimited number of variables and several operations, such as drive, emotion, and conditioning, resulting in high variability. As already mentioned, though, Skinner suggested that, with broad control over the variables relevant to the behavior at hand, the necessity that characterizes a reflex relation also is applicable to operant behavior.

As changes regarding the notion of the reflex occurred, the definition of the reflex as a correlation between stimulus and response started to assume variations in individual sequences. According to Skinner (1938), responses “are selected at random from the whole class that is, by circumstances which are independent of the conditions determining the rate” (p. 36). That is, variation was explained as a product of uncontrolled conditions, which interfere in the experiment. However, this variation could be eliminated by appealing to secondary laws.

Response differentiation, discussed by Skinner (1938), evinced how within-class variability, in the case of the operant, in the course of successive ‘elicitations’, is crucial for the occurrence of differential reinforcement relating to a particular property. Even though up to this point the term “selection” had not been used, this is the first time that implicitly the idea of complementarity between variation and a selection process -represented by reinforcement - appears: “Even where there is little variation, differential reinforcement necessarily involves extinction, and some strong responses are made available” (Skinner, 1938, p. 314). So, in this context, an alternative treatment of variation seems to be offered, because the discussion is no longer about eliminating variation; variation now is considered raw material for the reinforcing stimulus action, a treatment that is characteristic of a selectionist perspective. In this case, not only does reinforcement allow within-class variability, such variability is required to differentially reinforce certain responses.

Parallel to these nuances of a distinct interpretation of variation in Skinner’s system, the debate about secondary laws was predominantly a product of the 1930s.

In a subsequent chapter about response differentiation, Skinner (1938) returned to the concept of drive, another notion inferred to account for variability. He wrote:

The problem of drive arises because much of the behavior of an organism shows an apparent variability. A rat does not always respond to food placed before it, and a factor called its ‘hunger’ is invoked by way of explanation. The rat is said to eat only when it is hungry. It is because eating is not inevitable that we are led to hypothesize an internal state to which we may assign the variability. Where there is no variability, no state is needed. (Skinner, 1938, p. 341)

This excerpt illustrates how the absence of an inexorable relation between stimulus and response implies the necessity of inferring concepts to account for the variability found in some behavioral relations. Contrasting his approach to one based on the inference of metaphysical concepts, even though (a) regularity of the operant is not as obvious as it is for the reflex and (b) the variables responsible for differences in strength might not be easily identified and controlled, Skinner (1938) suggested that the change in reflex strength is an orderly process. In other words, variability can be described as a regular aspect and explained scientifically, preserving the reflex nature of behavioral relations, without risking losing the scientific status of his explanatory system.

As already pointed out, Skinner (1947/1999) emphasized probability as the basic data of his scientific system. Reflecting on this concept, its relation to the idea of variation is clear, since the reference to probability only makes sense if there is variability in the behavioral relation. Hence, the variable aspect of behavior seems to demand descriptions in probabilistic terms, and these descriptions imply the recognition of variations in this relation. Nevertheless, it was also mentioned that in the 1940s Skinner still was committed to the premise of complete determination of behavior, making it possible to suggest that the inference to probability results from the impossibility of controlling all the variables that determine the behavior. If this analysis is correct, even though Skinner did not mention variability in the 1940s, it is reasonable to assume that if it had been mentioned, the treatment of variability would follow the same line of reasoning that prevailed in the 1930s.

The only mention of the term ‘variation’ in the 1940s, from the texts analyzed, was in Walden II, in a discussion of the possibility of environmental control so carefully planned that it generates youngsters with highly stereotyped behavior inside the community. Considering such a possibility, Skinner (1948), through his character Frazier, explained that although the children in Walden II live in the same environment, their behavior is highly variable. Moreover, Skinner argued that it was a characteristic of this community to teach its members to respect and accept physical and behavioral limitations and differences. In that sense, variation seems to have been interpreted as a natural characteristic of community members, valued and expected by the group. So this explanatory perspective, more focused on the scope of human behavior, may have been a preliminary step in the elaboration of an alternative and selectionist treatment to behavioral variability.

It is necessary to emphasize, as well, that in Walden II, environmental control over individual behavior is a crucial feature of the community. That is why Castle’s questions about variability, spontaneity and freedom in the face of a highly controlled context shows how the recognition of a naturally variable aspect of behavior could, at this point, make way for traditional notions based on the conception of autonomous men, interpreted as free agency. Therefore, as suggested by Leão et al. (2016), Skinner, at that point, had found in the deterministic premise the only alternative to a scientific explanation. From Walden II it is not possible to conclude that variation had been predominantly interpreted as a complementary process to that of selection for explaining behavior, because variation was still linked to mentalist interpretations and backed by metaphysic connotations.

The idea of evolution based on three variation and selection histories is intrinsically dependent on the consolidation of the first two defining aspects of the selectionist mode, that is, the recognition of the selective role of consequences and an explanation based on the relation between variation and selection, previously analyzed. Despite the signs of the construction of these two initial characteristics in the 1930s and 1940s, there was no significant mention of natural selection and no direct reference to cultural selection in this period. From this, it seems evident that, at this early stage of Skinner’s system, the last characteristic of selection by consequences – cultural selection - was yet to be elaborated. However, some of his observations which refer to verbal behavior and to group or cultural level of analysis were selected to be discussed in this section, in order to demonstrate that some fundamental aspects for the construction of what will be later labeled the third level of selection at this point already are present in Skinner’s work.

In a discussion of the role of reinforcement as an operation that strengthens the operant, Skinner (1938) referred to verbal behavior, which he defined as that “part of behavior which is reinforced only through the mediation of another organism” (p. 116). This type of reference is relevant because the explicit notion of verbal behavior and the recognition of the role of the other in scope of such behavior will be indispensable in thinking about behavioral relations at the group level and about the status of individuals as social beings. The analysis of verbal behavior became better outlined in the 1940s, when Skinner (1945) presented an alternative proposal to traditional language theories and highlighted, as noted before, the role of the verbal community in the acquisition and maintenance of language.

Therefore, the importance of verbal behavior and the contingencies originated from a verbal community for explaining human behavior is already explicit. It is important to note that Skinner’s analysis is restricted to the individual level at this point, referring to the basic 3-term contingency, but with emphasis on the reinforcement offered by other’s actions. In addition, as noted in the preface to Verbal behavior (1957), the genesis of the book dates from the late 1930s, with a nearly finished version of the book in the mid-1940s, material that was the basis of his William James lectures at Harvard and used in courses taught by Skinner on the topic at several universities over these two decades. Thus, it is no surprise that there were analyses of verbal behavior in this initial period of Skinner’s career.

In 1947, in defense of the possibility of controlling human behavior, Skinner affirmed that genetic constitution, personal history, and social environment all have a role in determining behavior. As Micheletto (2016) pointed out, considering the link between these variables and natural selection, individual reinforcement history, and cultural contingencies, respectively, this kind of statement may suggest that behavior was already being perceived and understood by him as a product of three histories of variation and selection:

The genetic constitution of the individual and his personal history to date play a part in this determination. Beyond that, the control rests with the environment. The more important forces, moreover, are in the social environment, which is man-made. Human behavior is therefore largely under human control. (Skinner, 1947/1999, p. 319)

The analysis of social interactions and of the individual as member of a culture dominates Walden II, where Skinner (1948) defended the possibility of planning society through behavioral engineering. Even though the terms evolution and cultural selection were not yet explicit, there are strong signs of the notion of evolutionary processes in the development of cultures at this point of Skinner’s system. Evidence of this concern is Skinner’s (1948) highlighting of the importance of self-control of the individuals in the community, with a view to priority promotion of the well-being of society to the detriment of the immediate benefit for the individual. This does not mean, however, that Skinner’s system ignores the development of the individual’s well-being, but rather that the system aims to assure that individuals are committed to the group’s well-being. Otherwise, “the culture will eventually be replaced by competing cultures which work more efficiently” (p. 271). Therefore, by the late 1940s, Skinner considered cultural survival as a special value.

Furthermore, Skinner (1948) focused not only on how individual behavior directly influences the evolution of a culture, but also the fact that it is possible to intervene in the course of this evolutionary process through behavior technology. As he put it: “We can’t be satisfied with a static culture. There’s work to be done, if we’re to survive… We need a powerful science of behavior” (Skinner, 1948, pp. 290291). Beyond the community in Walden II, Skinner often pointed to the institution of family as “a medium for perpetuating a culture” (p. 138), in the sense of practices, customs and habits, implying that the family is the place to describe relations at the cultural level. Therefore, although the idea of behavior evolution based on three histories of variation and selection is not systematically established, there is clear evidence in this period of the notion of a selection process in the scope of culture and that transcends the individual level.