A Case for Observability

Abstract

Observability is a tricky concept that has been used by philosophers and scientists in an inconsistent and vague way. In this article a reformulation and operational analysis (as used by Skinner, 1945) of this concept is proposed and its implications are discussed. According to the view presented in this article, observation is defined as the act of making contact with a natural phenomenon and should not be conflated with observability, which is defined as the potential to make contact with a natural phenomenon. On the basis of our current faculties and tools, observability may be divided into four levels, labeled as (1) public, (2) private, (3) technology-enhanced, and (4) conceptual. Conceptual observability (typically referred to as interpretation) is especially important for scientific purposes, as long as it is informed by observations conducted at the other levels. Entities that fail to classify in those categories should be considered unobservable. It is further suggested that because all natural phenomena by definition lie within the observability spectrum, the notion of existence might be restated in terms of observability. An observability-based truth criterion is also proposed, according to which a statement may be considered true insofar it tacts (i.e., is controlled by) an observable event or series of events. Last, some implications of the present conceptualization of observability for putative psychological entities will be discussed.

The notion of observability has been a central aspect of several philosophical and scientific endeavors, while at the same time it has attracted some important controversies and criticisms. Early on, Karl Popper emphasized the importance of observability, relating it to testability (Popper, 2002/1935, pp. 84, 86). According to this view, entities may be divided into the ones that can be observed (i.e., intersubjectively tested) and those that cannot. However, as Maxwell (1962) argued, the line that distinguishes observable and unobservable entities is an arbitrary one. This is for the observation of some phenomenon pertains to the faculties of the observer: a specific event could or could not be observed depending on the tool used (e.g., the naked eye, a telescope, a microscope). Thus, Maxwell (1962) concluded that we have no solid grounds to determine whether a phenomenon is observable or not. A similar analysis was offered by Carnap (1966, pp. 225–228), who conceptualized observability as a continuum ranging from direct observations with one’s own senses to indirect measurements with the use of a state-of-the-art technology. Contrary to scientists who tend to accept a relatively loose criterion, philosophers, Carnap asserts, usually consider as observable those entities that can be observed directly or, at most, with the aid of simple technological means. In any case, the line that distinguishes the observable to the unobservable is admittedly arbitrary and vague.

On the other hand, the status of observability was defended and further developed by other philosophical schools, notably Bas van Fraassen’s constructive empiricism. Van Fraassen developed a system for the classification of entities as observable or unobservable, proposing the following definition: “X is observable if there are circumstances which are such that, if X is present to us under those circumstances, then we observe it” (van Fraassen, 1980, p. 16). Thus, entities can be said to be observable if they can be observed under specific circumstances, and unobservable if they cannot. However, it seems that van Fraassen overemphasized the importance of direct observations (i.e., those conducted with the naked eye), which resulted in several shortcomings in his system (for a relevant discussion, see Contessa, 2006). Indeed, a considerable part of the scientific activity is dedicated to the study of objects and events completely unobserved by the unaided eye (e.g., the neural activity, the subatomic world, and myriad stellar bodies). Further, as it will be shown below, the conceptualization of observability proposed herein significantly diverges from van Fraassen’s analysis, in particular with respect to the relation between observability and existence.

In the behavioral tradition, observability has played a prominent role, while evoking serious controversies. Some behaviorists have argued for the necessity that our subject matter be directly observable, whereas others have welcomed what they call private events in their analysis. J. B. Watson was the first behaviorist who attacked the method of introspection of the emotions and the states of mind, for they cannot be publicly observed and objectively studied (Watson, 1913). Nowadays, molar behaviorists follow a similar pathway, dismissing any talk with respect to private events, by virtue of their public unobservability and the inadequacy of introspection as a scientific method (e.g., Baum, 2011). In any case, the methodological and molar behaviorists’ objections to private events seem to be mostly of an epistemological, rather than a metaphysical, nature.

Contrary to the above-mentioned views, Skinner’s radical behaviorism made room for private events early on (see Skinner, 1945), advocating for their importance, as well as our ability to operationally (i.e., functionally) describe them. In his words,

[Radical behaviorism] consider[s] events taking place in the private world within the skin. It does not call these events unobservable, and it does not dismiss them as subjective. It simply questions the nature of the object observed and the reliability of the observations. (Skinner, 1974, p. 18)

However, it should be noted that although radical behaviorists emphasize the importance and scientific status of private events (e.g., thinking, remembering, feeling), they unswervingly reject mentalistic entities (e.g., Moore, 1990; Skinner, 1974). Mentalism (or cognitivism) is a widespread philosophical orientation, which posits the existence of internal entities (e.g., the mind, the language faculty, representations), usually with a presumed causal role on behavior (for a detailed analysis and definition of mentalism, see Moore, 1990). Radical behaviorists view those entities as unobservable, hypothetical constructs that add nothing to the explanation, prediction, and control of behavior.

Yet, Burgos and Killeen (2019) have recently raised some important concerns with respect to this issue. In particular, they advised against any attempts to distinguish private events and mentalistic concepts in terms of their putative observability or unobservability. This is for the “mind is ‘private’ because internal and, hence, no more or less publicly unobservable than the ‘private events’ some radical behaviorists propound” (Burgos & Killeen, 2019, p. 253). For these authors, we cannot repudiate mentalistic entities as inevitably unobservable, while accepting private events as scientifically legitimate. Burgos and Killeen suggest that given that both kinds of entities are unobservable to the public, radical behaviorists cannot criticize mentalism using observability-based arguments.

This brief introspection seems to reveal that there is little clarity and even less consensus regarding the concept of observability. Further, as Burgos (2021) convincingly showed, psychologists typically take the distinction of observable and unobservable entities for granted, without providing any clear definition of them; this seems to have resulted in the hampering of meaningful and beneficial discussions. The present article aims at providing a solution to this issue. In what follows, a reformulation and operational analysis (Skinner, 1945) of this notion will be articulated. It will be suggested that some of the problems and controversies discussed above have been caused by the conflation of the notions of observation and observability, which—as will be argued later on—should be viewed as distinct. Further, we will propose a system of describing observability (i.e., identifying and classifying the observable entities) and we will discuss the notion of the unobservable. We will subsequently examine whether observability could serve as an answer to the ontological question of what exists and as a truth criterion. Last, we will consider some implication of the notion of observability for mentalistic concepts, notably memory faculties.

Observation versus Observability

Does a falling tree in the forest make a sound when no one is there to listen? This is an old and tricky philosophical question that has bedeviled lots of thinkers over the years. Several different answers have been proposed based on the philosopher’s metaphysical commitments. Many will argue that the sound of the tree is real regardless of the presence or the absence of an observer. Others might respond that unless perceived by someone, an event does not exist. Some others would object that this question is irrelevant or pointless, because it makes no practical difference whether a lonely tree makes a sound or not.

It seems to me that much of the dispute stems from the fact that this question conflates two different concepts. The stimulation we, organisms, make contact with and call a sound is a vibration of molecules that are transmitted through the air (or some other medium) and hit the sensory receptors of our eardrum. In the case of a lonely falling tree, the two actual questions to be answered are (1) whether the air molecules hit the eardrum of an observer when the observer is not present, and (2) whether an observer with the appropriate tools (e.g., intact hearing, a sound meter) could potentially witness this vibration of the air molecules. The two different concepts that need to be clarified are observation and observability. In what follows we will attempt an operational definition of those two concepts. To be sure, this kind of operationism is meant in the Skinnerian sense (Skinner, 1945) of an analysis of the conditions that may evoke the verbal responses observation and observability.

Observation refers to a specific event, an act that has been already or is currently being conducted; it is the act of making contact with a natural phenomenon. Observation is a single event with spatio-temporal characteristics, an act with well-specified starting and ending points. A person might or might not observe (i.e., make contact with) a specific fall of a specific tree at a specific time. On the other hand, I propose that we define observability as the potential to make contact with some phenomenon. In this sense, observability is not a specific act, nor is it relevant to a specific instance of an event; it rather refers to the potential to observe a natural phenomenon. Thus, the vibration of the air molecules by the falling of a lonely tree is a phenomenon perfectly observable (i.e., it can be observed given the appropriate circumstances and tools), even though not observed by anyone.

Observability is sometimes taken to be an intrinsic property of the phenomena under study. That is, we may be tempted to say that it is the very essence of some entity that renders it observable or unobservable. However, this view would be an essentialist description of nature—a bad practice for biological and behavioral sciences (Mayr, 2002, pp. 80–84; Palmer & Donahoe, 1992). Further, observability presupposes a subject who is to witness an event, and it should be defined in terms of this subject (Palmer, 2003, 2009). Thus, the question of whether a falling tree “really” makes a sound is actually a question of whether a person with the appropriate faculties and tools (e.g., intact hearing, a sound meter) and in the right circumstances (e.g., when the tree falls, when the background noise is below a specific level) could potentially make contact with the phenomenon.

Some Characteristics of Observability

In the previous section, we attempted to operationally define and differentiate the notions of observation and observability. Observability refers to the potential of an observer to make contact with a natural phenomenon; observation refers to whether an observer has actually made contact with a natural phenomenon. Now, let us take a closer look at the concept of observability, in particular the criteria to classify an entity as observable and the “absolute” nature of it.

We may say that a stimulus or event has been observed if someone has made contact with it. Yet what is the criterion to call a phenomenon observable? In other words, under what circumstances may we categorize an entity as observable? It seems to me that the act of observation should be accepted as the strongest criterion to predicate observability; that is, we should always call an event observable if it has already been observed. For example, we can call the Milky Way observable, simply because in the course of the history at least one person has actually witnessed it. Yet, this criterion cannot be the only one. Recall that we defined observability as the potential to make contact with an object or event. This potential should also apply to objects and events that have not yet been observed. That is to say, even though the actual observation is the strongest criterion for acknowledging observability, the lack of observation does not in and of itself preclude observability. For example, black holes had not been observed until very recently; however, we now have a beautiful photography of a black hole. Undoubtedly, there are myriad natural phenomena yet to be discovered (i.e., observed), but this fact does not render them unobservable. Whether a phenomenon or object is currently unobserved has nothing to do with our potential to witness it.

Two examples of entities that have never been observed are the earth’s core and the fairies. Indeed, we do not have yet any means of approaching and directly interacting with the earth’s core, and there is no evidence whatsoever that fairies are flying around us.1 Should both of those entities be called unobservable because they are unobserved? Certainly not. In fact, scientists have provided a very plausible account regarding the temperature, elements, and “behavior” of the earth’s core based on indirect knowledge, such as seismic activity or the elements that already exist on Earth. To the contrary, fairies bubble up out of nothing; that is, any talk about them is not based on directly observable data. Thus, it seems that a second criterion to acknowledge observability should be that our inferences about currently unobserved entities are an extension of already observable phenomena (Palmer, 2003, 2009). We will further examine and expand on this argument below. For the moment, let us turn to another issue: the absolute nature of observability.

Recall that we defined observation as the act of making contact with an event or object and observability as the potential to make contact with an event or object. However, the potential to observe does not necessarily ensure the act of observation, unless a certain set of circumstances are met. The Milky Way can indeed be observed, but not at any moment; if we attempt to see it in the daylight or during a cloudy night, it will go unnoticed. Such a rationale has been used to indicate that observability of some phenomenon is not constant, but rather changes in accord with the faculties and tools of an observer (e.g., Maxwell, 1962; Palmer, 2003, 2009). For example, arguing against a dualist view of overt and covert behaviors, Palmer (2009) wrote: “[Observability] is determined by the vantage point of the observer and is not a property of the behavior itself” (p. 12). However, according to the current conceptualization, it is the act of observation, not observability, that is tied to an observer’s vantage point. The fact that a phenomenon can be observed in some circumstances and cannot be observed in others is a matter of observation. To the contrary, even one observation of some object or event is sufficient to indicate its observability (i.e., our potential to make contact with it).

The circumstances in which a phenomenon cannot be witnessed do not render it less observable, but rather less observed. Daylight does not compromise our potential to make contact with the Milky Way; it merely inhibits our act of observation. What has changed since the first photograph of a black hole is our act of observing it, not our potential to observe it. In other words, the photograph did not increase the observability of black holes because our potential to make contact with them was always there regardless of our being aware of it. What has changed is the actual contact with them. Observability, as defined here, should be considered an absolute (as opposed to a “protean”; see Burgos & Killeen, 2019) concept: an entity either is or is not observable. Stated another way, the only two possible answers to the question “Is X observable (i.e., can we potentially make contact with it)” should be “Yes” or “No.” The circumstances under which we can contact X are of great importance, but they pertain to observation rather than observability.

Yet, how about phenomena that are currently unobserved and elusive even to our imagination? Viruses and bacteria were once completely unobserved, while our ancestors lacked the necessary tools and conceptual schemes to infer their existence. In the same way, it seems safe to assume that we are currently ignorant of the existence of numerous natural phenomena that might or might not be observed in the future. Does this fact compromise the absolute nature of their observability? I believe not. The absolute nature of observability refers to our potential of making contact with a natural phenomenon. Whether or not we are in a position to verbally classify a phenomenon as observable, does not guarantee or preclude its actual observability. Our ancestors might have laughed at the idea of the existence of tiny living creatures that make them sick and categorize such creatures as nonexistent (i.e., unobservable). However, our ancestors’ failing to verbally acknowledge the potential of making contact with microorganisms should not be conflated with the actual observability of microorganisms. In other words, although we can classify a phenomenon as observable or unobservable, we need to remember that this classification might be flawed. But a natural phenomenon would always be observable if we could potentially make contact with it and always unobservable if we could not, hence the absolute nature of its observability.

Consider the current status of fairies in Western societies: they are believed to be unobservable, fictional entities. Now, suppose that new data arise indicating that fairies are in fact observable. As counterintuitive as this might seem, we would have to acknowledge fairies as “real.” However, this would not mean that the observability of the fairies has changed. To the contrary, fairies would have always been observable, but we would simply have failed to acknowledge this fact. What we say and believe about the observability of some phenomenon should be viewed as verbal behavior and nothing more. Our classifications may prove to be wrong, but this does not affect the observability of an entity. Phenomena can be either observable or unobservable, and there are no gray areas. In other words, observability (i.e., our potential to contact a phenomenon) should be considered an absolute concept. This fact seems to create a problem because we cannot be absolutely certain that any putative entity is in fact unobservable. But this should be viewed as a flaw of our perceptual and verbal behavior, not of the phenomena themselves. Classifying phenomena as observable or unobservable is necessary for scientific purposes, despite the possibility that our classifications may be wrong.

The Four Levels of Observability

As already discussed, observability cannot be regarded as an intrinsic aspect of a phenomenon; it is, rather, a verbal utterance that indicates our potential to make contact with an entity or event. Thus, the question “Is X observable?” is equal to asking, “Could we potentially observe X, given the appropriate circumstances and tools?” A blind person cannot see the falling tree, whereas a deaf person cannot hear it. A person in love can notice their internal sympathetic arousal, but the others cannot. The astronomer is able to detect with a telescope a most distant star that cannot be observed with the naked eye. However, as we pointed out, all those phenomena are nevertheless observable; that is, there are ways for at least some people to make contact with them. In this sense, it is not observability, but the act of observation that is tied to the tools and the vantage point of the observer (c.f. Palmer, 2003, 2009). Observability, as defined here, should be considered an absolute concept. The tools and the vantage point of a witness do not render a phenomenon more or less observable; they only make the act of observation more or less possible.

We may be able to better talk about observability if we break it down into levels in terms of our current faculties and tools. To be sure, these levels are not “real” categories with discrete boundaries nor are they to be found in nature. Rather, they are made up for the sake of convenience when discussing the notion of observability. Even more important, whether a phenomenon is observable (i.e., whether some could potentially make contact with it) is not dependent upon the level at which the phenomenon is classified. What those levels indicate is what person, with which tools, and from which vantage point might or might not be able to observe that phenomenon.

The first level might be called public observability. We may say that a phenomenon has been publicly observed if two or more people have directly witnessed it with the naked eye. As a result, this phenomenon is publicly observable; that is, two or more people could potentially make direct contact with it under specific circumstances. Direct observations in the laboratory or in naturalistic settings indicate public observability of an event. We might say that what a person in an experiment, in a clinic, home, community, etc., says or does has been publicly observed, hence it is a publicly observable behavior. The falling or the sound of a tree are publicly observable phenomena, regardless of whether a specific fall has or has not been observed by anyone. To wit, everything that has been observed with the naked eye (or heard with the “naked ear,” or smelled with the “naked nose,” etc.) can claim the title of the publicly observable.

The public level of observability is to be contrasted to the private one. We may talk about private observability when only one person is able to observe an event. In this case, there are certain limitations in the current ability and tools of the community to witness some phenomenon. A person who speaks only to themselves being exposed to their own speech or a tree that falls in front of only one person, are phenomena privately observed. However, such events are publicly observable, because they can be potentially witnessed by more than one people. There are, though, other events that are, if anything, privately observable. Suppose that a person is thinking of a problem or is imagining a picture or is remembering a summer vacation. Think of a person who is feeling depressed or anxious or happy; imagine of a person who is seeing a photo or is listening to a song or is smelling the roses. Such activity (feeling, sensing, thinking, imagining, remembering, etc.) can be only observed by the person engaging in it. No other person can tell whether they are seeing or listening or imagining or remembering. The only one who can observe their own within-the-skin behaviors is the behaving person themselves; hence those behaviors are only privately observable.

Radical behaviorists hold that privately observable behavioral events have a status no different from publicly observable ones. The only difference between overt and covert responses is the locus in which the latter are happening (i.e., within the skin of a person) and the inability of more than one people to witness them. Indeed, the critical characteristic of radical behaviorism that distinguishes it from other behaviorist systems, is this very consideration of privately observable events (e.g., Skinner, 1945, 1974; Day, 1983; Moore, 1984, 1990, 2009; Palmer, 2003, 2009, 2011, among many others). This view is justifiable by the present formulation, for observability is an absolute concept: a phenomenon either is or is not observable. There is no commitment with respect to the individual who makes contact with the phenomenon or its accessibility by the community. Further, regardless of the locus in which a natural event resides, it always remains a natural event (and an observable one). This way, privately observable phenomena are “accessible only to one person, but nevertheless physical and material” (Moore, 1990, p. 24). That is to say, we have no reason to consider that the within- and the outside-the-skin behaviors obey different natural laws, or that they are of a different metaphysical status. The only fact that should be acknowledged is the limited access of the community to the privately observable phenomena (Skinner, 1957, p. 130).

However, there are a large number of events taking place inside or outside the skin of an organism, which cannot be observed by them. Instead, they are detectable with the aid of technological advancements. One example of this third level of technology-enhanced observability is the activity of the nervous system, which can be observed only with the aid of advanced technology. Other instances of technology-enhanced observations are studies of the microorganisms, observations of the subatomic world and the identification of numerous stellar bodies, to name a few. Because more than one person can potentially observe a phenomenon with the aid of the technology, technology-enhanced observability might be viewed as a subcategory of public observability. That is to say, technology-enhanced observations are direct observations of a phenomenon, rendering it publicly observable. A microscope amplifies the capacities of the naked eye and a hearing aid those of the “naked ear.” As long as our tools are reliable, valid, and appropriately calibrated, it is safe to consider them a legitimate means to make direct contact with (i.e., directly observe) a natural phenomenon.

The last level of observability is a tricky one. It cannot be denied that there are myriad phenomena, relations, and objects that currently remain unobserved. However, we can still make some contact with them using a conceptual scheme (i.e., an interpretation). In other words, we could provide a more or less plausible account for currently unobserved phenomena or for the currently unobserved relations between events—this fourth level being called conceptual observability. A conceptually observable phenomenon is by definition currently unobserved at the other three levels. That is, we cannot witness it yet, but we can nevertheless make some contact with it using our own verbal and perceptual behavior. To speak of such phenomena entails much uncertainty and our interpretations should be viewed only as tentative hypotheses (Palmer, 2003, 2009). In this sense, the conceptual level of observability is the most dubious one, because any interpretation, however plausible, could be proved to be wrong. Further, we conceptually describe an event with the anticipation that eventually we will be able to observe it at some of the other three levels. To provide a historical example, when Darwin encountered an exceptional orchid, the nectar of which was found at the end of an 11-in spur, he proposed the existence of a moth with an 11-in proboscis, which would necessary to fertilize the orchid. For decades, such a moth was only observed at the conceptual level, until it was videotaped in 1992—hence, it can be now called publicly observable.

Conceptual Observability and Unobservable Entities

At this point, we should examine the tricky part of the notion of conceptual observability: it does not come for free. Indeed, unless somehow restricted, we could possibly generate as many hypotheses as we please. For example, one might contend that the goblins fertilize Darwin’s orchid with a secret potion, that God blessed it with a miraculous ability of self-fertilization, and so on. To avoid such an unwarranted proliferation of speculations, which would be highly misleading and in contact with nothing, we need to establish valid criteria in order to evaluate which interpretations may be accepted and which ones may not. For example, many cognitive scientists maintain that a person is able to remember things past, because they have been stored in the memory faculty (e.g., Anderson, 2014, pp. 124–126). A radical behavioral account would view remembering as a result of stimulus control (e.g., Donahoe & Palmer, 2004, pp. 330–335; Palmer, 1991). In fact, no one has ever directly observed either some memory storage or the covert links of the operant chain that presumably ends with the production of the stimulus of interest. Countless other hypotheses could be possibly generated in an effort to account for such an intriguing and complex phenomenon. Do all those hypotheses help us to make an even contact with the phenomenon? How could we choose among them?

It has been proposed that the most valid criterion for the evaluation of an interpretation is whether it has been guided by direct or technology-enhanced observations (Donahoe & Palmer, 2004, pp. 125–129). In other words, we may accept conceptual observations (i.e., interpretations) as a valuable and scientifically legitimate practice as long as they originate from directly observable data. This is because, if we use actual observations as a starting point of conceptual extensions, there are more chances that we will make an adequate contact with the phenomenon under study. Following the hypothetico-deductive approach, for example, we could generate countless hypotheses for complex phenomena, but none of them would ever be an extension of directly observable data (for a critique of hypothetico-deductive approaches in psychology, see Chiesa, 1994, pp. 49–52). On the contrary, the process of stimulus control has been extensively observed with the naked eye, as well as with the aid of our technology. If, as radical behaviorists hold, overt and covert behaviors obey similar laws, a stimulus control interpretation of remembering would be acceptable by the present system. And, although it is true that an interpretation (i.e., a conceptual observation) informs the most doubtful level of observability or that it can be readily displaced by some better proposal, it is much needed in order to account for yet unobserved events. As long as an interpretation is an extension of the observations that have been conducted at the other levels of observability, we may tentatively rely on it, instead of adopting another position that would diverge from the actual data or avoiding to study a phenomenon altogether.

With respect to concepts formulated a priori rather than as an extension of observable data, we might say that they are unobservable, or simply fictional. In other words, a posited entity that does not stem from observable data should be considered unobservable, as plausible as it may seem to a layperson’s or even a scientist’s eye. As Skinner showed early on, adopting an everyday terminology may be convenient, but it may not point to the actual processes in effect (e.g., Skinner, 1957, p. 45; see also Skinner, 1945). For a phenomenon to warrant the title of the conceptually observable, it is implied that it could potentially be observed if or when the scientist gains access to the right tools. Again, conceptually observable phenomena should be considered only potentially observable, because we are not able to make actual contact with them yet and we may never be able to. However, tolerating such an uncertainty seems a better scientific practice compared to hypothesizing putative entities that do not stem from actual data.

Of course, behavior analysis is not the only field that resorts to conceptual observations of complex phenomena. Many other sciences have provided successful interpretations of currently unobserved phenomena, based on direct observations at the public and technology-enhanced levels. As we already noted, no one has ever directly observed the earth’s core, but geologists have formulated adequate interpretations rendering the earth’s core a conceptually observable object. Astronomers provide plausible accounts for the elements the outer planets consist of, based on extensions of technology-enhanced observations established in much more simple conditions. Darwin’s moth or the black hole were unobserved at first, but scientists had good reasons to posit their existence and conceptually describe them. And, indeed, we now have a photo of a black hole and a video of a moth with an 11-in proboscis. The common feature of those conceptual observations is that they are informed by data directly observed. A divine charisma of an orchid or the fairies or a memory faculty inside our head are not just currently unobserved (like the black hole or Darwin’s moth were once), but rather unobservable whatsoever. And, although it is true that a conceptual observation might prove to be inadequate (because some other conceptual observation may permit us a better contact with the phenomenon of interest), it is a totally legitimate and useful scientific practice. On the other hand, the postulation of a priori models and conceptual schemes seems to be a futile practice, for there are much lower chances that they establish contact with some phenomenon. As previously discussed, the classification of some entity as unobservable cannot be definitive, because new data may in fact indicate its observability. But, until such data accumulate and based on the current state of affairs, we can accept our current classification as an adequate guide for our scientific enterprises.

In the above discussion, we analyzed the notion of the unobservable, equating it to “fictional.” However, we should explicitly examine and elaborate on the relation between observability and existence, as well as unobservability and nonexistence. Do only the observable phenomena “really exist,” or observability and existence should be treated as separate concepts? These are questions we will be dealing with in the next section. In addition, we will examine whether the observability argument provided here can serve as a truth criterion.

Observability, Reality, and Existence

Is there a real world independent of our own perceptual and verbal behavior? As we discussed using the falling tree paradigm, the question “what is really there” may be much vague and misleading, because it may refer to many different things. A much more detailed and operational (Skinner, 1945) approach is needed in order to make sense of it. Given the close relationship between radical behaviorism and pragmatism in the epistemological field, it should be mentioned that pragmatists prefer to be silent or agnostic regarding the possibility of an independent reality, with the exception of Rorty’s system (see, Schoneberger, 2016). In any case, most pragmatists who have addressed these ontological questions argue for our inability to access such a purportedly autonomous reality, and therefore they prefer to plead their agnosticism or to dismiss ontological questions altogether (e.g., Barnes-Holmes, 2000). Such a position seems to be justifiable in terms of our observability argument. One might predicate that we can never observe some independent reality, because, once observed, it is not independent anymore. That is to say, we should have no means of witnessing or talking about something that is independent of us. However, the question of whether something exists remains unanswered.

Can observability shed some light to this issue? Before we discuss this possibility, let us review the position taken by Van Fraassen, who explicitly distinguished observability and existence. His system permitted entities to be unobservable but existent, as well as observable and nonexistent. In his words, “the term ‘observable’ classifies putative entities, and has logically nothing to do with existence” (Van Fraassen, 1980, p. 18), and “[a] flying horse is observable—that is why we are so sure that there aren't any” (p. 15). Stated another way, for the father of constructive empiricism, observable entities are those that we would be able to witness with our unaided senses; whether they exist or not is completely irrelevant.

As the reader may have noticed, the present conceptualization of observability significantly diverges from Van Fraassen’s. The present view implies that we could potentially make contact with virtually any object, phenomenon, or event in the universe. As discussed above, observability does not depend only on the act of observation or, if you will, the current lack of observation does not by itself preclude observability. For example, there are phenomena conceptually observable to us, with the caveats that those conceptual schemes (i.e., interpretations) refer to potentially observable phenomena and that they should be informed by actual observations. Further, there are virtually infinite unknown objects and events yet to be discovered (i.e., observed); but this fact says nothing about our potential to make contact with them. Thus, if any phenomenon in the universe is potentially observable, and if observable phenomena are only those that we could potentially make contact with, another possibility arises. In particular, it might be possible to restate the vague and problematic question of whether some phenomenon “really” exists, as whether it is observable. Following this view, the falling of a tree would be “real” only in the sense that it is observable. The fairies would not exist merely in the sense that they are unobservable whatsoever. To say that an entity exists may be simply a crude way of saying that we could potentially make contact with it. Thus, instead of adopting a criterion based on a supposedly “independent” reality, we might prefer to talk about natural phenomena in terms of observability.

Following this rationale, to be unobservable would mean to not exist. Van Fraassen’s flying horse is not considered here as an observable entity, because it has not been observed in any circumstances and because there is no empirical basis to assume that one day it could be. This flying horse is unobservable whatsoever—hence nonexistent. According to the current conceptualization, only phenomena that lie within the continuum of observability (regardless of the level they are classified at) may be said to be “real.” Such a rationale might be followed for the purpose of clarifying debates relevant to putative psychological entities that are unobserved with the naked eye. In those cases, we could examine whether a putative entity may be observable at another level or rather unobservable, hence nonexistent.

An Observability-Based Truth Criterion

Another important question is relevant to the “truth” and the criteria to assert that some statement or belief is true. Most metaphysical realists tend to endorse a correspondence-based truth criterion (David, 2015); that is to say, a statement is considered true to the extent that it corresponds to the independent reality that allegedly exists regardless of our own perceptual and verbal behavior. However, correspondence theories of truth are to be rejected in terms of observability, for the same reason we argued against an absolute reality; that is, independence and observability are incompatible concepts. Because we would be unable to make any contact with such an alleged reality, we could not estimate the degree that a statement or belief corresponds to it.

On the other hand, several behaviorists have proposed and defended a pragmatic truth criterion (e.g., Barnes-Holmes, 2000; Moore, 2011). This is the criterion of successful working, typically assessed in terms of prediction and control. Following this rationale, we are said that a statement or a theory is true insofar it promotes effective prediction and control of some phenomenon. However, it seems to me that these criteria, however helpful, are not adequate to estimate the truth or the scientific value of some statement or model. As we already discussed, cognitive psychologists consider remembering as a manifestation of the function of some internal memory faculty of the mind. The various cognitive models of memory have indeed produced a vast body of experiments and practical applications, unlike the behavior analytic views that interpret remembering as an instance of stimulus control. In the clinical domain, cognitive, psychoanalytic, and client-centered theoretical models have resulted in a considerable number of evidence-based psychotherapies. Such applications are indeed effective, but presumably for reasons other than those posited by the theories they stem from (Chiesa, 1994, p. 49). Based on a pragmatic truth criterion of prediction and control, we need to accept the truth of, for example, the psychoanalytic theory that informs the evidence-based brief psychodynamic therapy for depression. This is a peculiar position for a behavior analyst to take.

An observability-based truth criterion might be stated as follows: a statement may be considered true insofar it names an observable event. Consistent with the Skinnerian terminology (Skinner, 1957), we might call a statement true when it tacts (i.e., it is controlled by) an observable stimulus. This stimulus or event or series of events might be classified at any point across the observability continuum. To say that the statement “A red book is on my bed” is true, is to accept that it tacts an observable set of phenomena and the relations between them. A client’s answer to the therapist “I’m really sad” reflects the truth if this statement is evoked by a relevant privately observed feeling. The doctor who informs us “Your blood pressure is low” is telling the truth if (1) they accurately tact the results of the blood test, and (2) the technological tools they used are accurate and valid. The interpretation that the cave art of Altamira dates back to 35,000 years ago, is true to the extent that it is informed by actual data at the other levels of observability. To recapitulate, an observability-based truth criterion would be whether a statement or a belief tacts (i.e., is controlled by) an observable event or the relations between observable events.

Yet, what about “hallucinating” people who avouch that they can see the fairies or that they once fought alongside Napoleon Bonaparte? Is it not true that they have experienced what they claim to? In order to answer questions like these, note that a tact (in the lay vocabulary: naming, labeling, describing, etc.) is always controlled by an antecedent nonverbal object, stimulus, or stimulus property (Skinner, 1957, pp. 81–82). Thus, as long as we classify the above-mentioned statements as tacts and not something else (e.g., a mere request, or mand, for attention), we need to consider the event that evoked them. This way, it is actually true that the first person has observed the fairies and the second one was in the Napoleon’s camp. However, we have every reason to assume that the stimuli of interest (i.e., the fairies and the participation in the Napoleonic wars) were merely the products of the perceptual behavior of those people. Stated another way, even though it is not true to say that the fairies are flying around us, it may be true that fairies are privately observable as the automatic product of the within-the-skin behavior of a hallucinating person. Likewise, based on direct evidence from other sciences (e.g., Napoleonic wars ended in 1815, our species has a limited life span), it is not true to say that a contemporary human fought alongside Napoleon, unless as a result of their own perceptual behavior.

To reiterate, everything that has a “real” (i.e., observable) effect upon a person’s behavior is indeed a “real” (i.e., observable) event. Our task, therefore, is to detect the locus in which this event has happened, that is, within the skin of a person or outside it. Extending this rationale to putative psychological entities, we may be able to shed some light to several long-lasting controversies. What implications may the present formulation of observability have for mentalistic entities like memory faculties, representations, or the mind? Is it possible to consider them as conceptually observable or should they be viewed as simply as a word or a metaphorical extension as radical behaviorists contend (Skinner, 1957, pp. 92–99; Moore, 1990).

Observability-Based Implications for Mentalism

Mentalism is a philosophical position deeply rooted in the Western cultures and reflected not only in our everyday discourse, but in mainstream psychology, too. Moore (1990) defined mentalism as

. . . [a]n orientation to the study of behavior, which holds that a unique, a necessary, and the primary contribution to the causal explanation of behavior consists in proposing various internal acts, states, mechanisms or processes, presumed to be operating in neural, conceptual, or psychic dimensions. (p. 20)

In other words, mentalistic views postulate the existence and advocate the causal role of internal entities of an either physical or a nonphysical essence. This position has been historically criticized by radical behaviorists on epistemological and metaphysical grounds. For example, it has been argued that mental entities have distracted psychologists from the effective study of environment–behavior relations (Skinner, 1974, pp. 15–16; Moore, 1990). On the other hand, metaphysical objections have pointed to the close relation between mentalism and dualism (Skinner, 1974, p. 10; Moore, 1980, 2011).

However, as we discussed earlier, it is not rare for mentalistic concepts to yield a rich research program and effective practical applications. Further, it seems that not all mentalistic branches are prone to the dualism criticism—a prominent example being the mind–brain identity theory (Burgos, 2016; Burgos & Killeen, 2019). On the other hand, the concept of observability as articulated here and some of its implications may be relevant to the debate about mentalistic entities. However, because Burgos and Killeen (2019) have provided sound arguments against observability-based objections to mentalism, let us begin by examining and discussing those arguments. First, Burgos and Killeen hold that radical behaviorists should not criticize mentalistic entities in terms of observability, for those entities are no less observable compared to radical behaviorists’ private events. Why would radical behaviorists accept their putative private events while rejecting mentalistic concepts, because none of them are directly detectable? Expanding on this argument, Burgos and Killeen (2019) are polemical to the views that overemphasize the importance of observations with the naked eye, objecting that science has historically relied on technological amplification and theoretical constructs in the study of most important natural phenomena.

Yet, it seems that the notion of observability as articulated here resolves many of the problems Burgos and Killeen (2019) have raised. Recall that, according to the present formulation, observability was divided into four levels: public, private, technology-enhanced, and conceptual. The present view corroborates Burgos and Killeen’s remark that acceptable scientific activity is not to be restricted only to the public domain. Observability is defined as the potential to make contact with an event and this contact can also be satisfied with private observations, technological amplification, and potentially with conceptual schemes. As long as our interpretations are careful extensions of directly or technologically observable data, the possibility of making actual contact with the phenomenon of interest increases.

However, what implications may our formulation of observability have for mentalistic entities? In modern cognitive science, observations from neuroscience have been largely used as an evidence for the support of various mentalistic models. For example, it has been suggested that data obtained from neuroimaging studies, however inconclusive, seem to support the distinction between a short-term memory (where information is held for relatively short time-frames) and a long-term memory (where information is encoded, stored, and obtained from; Jonides et al., 2008). One argument might be that memory architecture and processes can be considered conceptually observable phenomena, because relevant accounts have been based on technology-enhanced data.

In fact, there have been several behaviorist analyses as well that link covert behaviors (e.g., remembering) to neuroscientific data (e.g., Donahoe & Palmer, 2004; Ortu & Cihon, 2019; Schlinger, 2015). According to those views, by using neuroimaging techniques, we observe the activity of the brain and we are indeed able to find neural correlates to overt or covert behavioral processes. However, it seems that many mentalistic theories go one step further by hypothesizing that the neural data indicate the existence of internal entities different from the covert processes radical behaviorists acknowledge. It is one thing to recognize that, when people are performing everyday physical actions, an increased activity of the cortex is observed as well, and a different thing to postulate that this happens because “implicit memories are stored in the cortex” (Anderson, 2014, p. 177). The latter assertion contains many unobservable inferences that do not necessarily follow from the technology-enhanced observations provided by neuroscience. Likewise, a person who is reciting a word-list may be able to observe this covert behavior, rendering it a privately observable phenomenon; however, they cannot observe any memory faculty of the mind, which allegedly makes this behavior possible.

According to radical behaviorism, we should conceptually analyze the act of imagining, say, one’s summer vacation, in the sense of “seeing in the absence of the thing seen” (Skinner, 1974, pp. 91–95). Such an analysis can be achieved by extending our well-established behavioral principles to the publicly unobserved phenomena (Palmer, 2003, 2009). However, it does not follow that the act of seeing in the absence of the thing seen has been made possible because of the storage and retrieval of representations in and from one’s memory faculty. And no matter how many neuroscientific data may link different brain structures to different remembering behaviors, the putative encoding, storage, and retrieval of information from memory faculties seems to be an unobservable entity.

Conclusions

To recapitulate, the present article intended to discuss the notion of observability under a behavioral viewpoint. It was first showed that observability has caused several problems to philosophers and it was suggested that most of those problems stem from a flawed conceptualization and definition of it. Observability was then operationally defined as the potential to make contact with a phenomenon—an absolute concept; this contact could be of a public, private, technology-enhanced, or even a conceptual sort. However, we need to set up some criteria for a conceptual scheme (i.e., interpretation) to be acceptable. Thus, echoing radical behaviorists’ arguments (e.g., Donahoe & Palmer, 2004), we endorsed the view that any conceptual observation needs to be informed by already conducted observations at the other levels. Thus, for a phenomenon to be regarded as conceptually observable, it should be an extension of actual observations. This is for, viewing a currently unobserved phenomenon as an extension of what we have already observed permits us much higher chances of making contact with the phenomenon of interest, compared to the practice of generating countless unjustified hypotheses.2 However, we need to remember that our interpretations (i.e., conceptual schemes) can be corrected or displaced following new evidence. This should not be considered a problem, but rather a legitimate and useful scientific practice.

Another objective of the present article was to examine some of the implications of observability, as defined in this article. Thus, it was proposed that we can talk about natural phenomena only in terms of observability, because we are by definition unable to make contact with some “independent” reality. In other words, because all the “existent” entities in nature can be potentially observed (i.e., we could make contact with them, given the right tools and circumstances), it might be justified to redefine existence or reality in terms of observability. In addition, an observability-based truth criterion was proposed: a statement or belief may be considered true insofar it tacts (i.e., is controlled by) a set of observable phenomena and/or the relations between them. Last, some implications for putative psychological entities were examined. According to the present view, it seems that some mentalistic concepts, such as memory faculties, fail to categorize as observable phenomena.

The present formulation of observability may help to resolve some of the problems Burgos (2021) identified with respect to the observable–unobservable distinction. A clear and coherent conceptualization of observability is of potential usefulness to those who wish to talk about observable and unobservable entities, while providing a possible empiricist answer to the ontological question of what exists. In the scientific field, the implications of observability may be even more interesting. For example, the present formulation allows behavioral scientists to consider private events (e.g., thinking, feeling, imagining) and technology-enhanced observations (e.g., studies of the nervous system) as something not fundamentally different to observations conducted with the naked eye. It also legitimizes scientific interpretation as long as it meets specific criteria. Last, the notion of observability may be used in order to clarify debates about the “existence” of putative psychological entities, thus guiding scientific activity to potentially useful paths.

To be sure, the present formulation and its implications are nothing but a conceptual observation of natural events, the relations among them, and the verbal behavior we may use. Thus, it might be the case that we can improve on the conceptual observations provided here in an effort towards a better contact with the natural phenomena. In any case, it is my hope that the present article will add to a much-needed discussion about the challenging and intriguing notion of observability.

Declarations

Compliance with Ethical Standards

The author declares that he has no conflicts of interest