Abstract
We argue that the question, “Should psychology follow the methods and principles of the natural sciences?” is not one that should be answered by theoretical psychologists or metascientists; rather, we implore psychological researchers themselves to heed Wittgenstein’s observation that a preoccupation with method and principles risks overlooking important conceptual issues, the clarification of which are necessarily antecedent to consideration of empirical activities. We examine potential conceptual problems that arise when psychological researchers attempt to follow natural science methods and principles without first considering the concepts that are relevant to their scientific pursuits. Drawing primarily from the works of Hacker, Lamiell, and Maraun, we argue against the dogmatic following of any methods or sets of principles. Instead, we posit that natural science methods and principles should be considered on a case-by-case basis after relevant psychological concepts have been carefully considered and empirical investigation has been deemed an appropriate path forward.
Since the inception of psychology, psychological researchers have appeared to grapple with their intellectual identities (see Goertzen, 2008; Teo, 2005). Given this longstanding struggle, it is unlikely that the question, “Should psychology follow the methods and principles of the natural sciences?” will be resolved in this special issue, and, in our view, this is perfectly fine. We approach this debate by drawing a distinction between conceptual matters and empirical matters. We argue that psychologists should focus first on carefully considering the meanings of concepts that they employ in determining their research questions. Although we argue that psychologists should strive for greater conceptual clarity, we do not believe that it is our job to tell psychologists how they should or should not carry out their empirical program of research. What we do feel strongly about is the persistence of conceptual confusions in psychological literature, often overshadowed by psychological researchers’ (both quantitative and qualitative) preoccupation with methods. Mainstream research in psychology is often critiqued (especially by theoretical psychologists and qualitative researchers, ourselves included) for its mimicking of the natural sciences, particularly its obsession with quantitative methods (see Tafreshi et al., 2016). However, it is our view that although psychology’s infatuation with natural science principles and methods can be problematic, the fundamental problem for psychology is the lack of care and attention paid to conceptual issues that, in turn, lead researchers down a path of conceptual confusion (Wittgenstein, 1953). No amount of examination of methods and/or principles will resolve such confusion.
There is a reciprocal relationship between psychology’s infatuation with science and the conceptual confusion that reveals itself in psychological science discourse. The attempt to emulate the sciences contributes to and exacerbates such confusion. However, it would be a mistake to assume that if psychology were to just accept its nonscience status such problems would be resolved, or that adopting alternative principles or novel methodological approaches would ensure careful thought around the use of concepts. Conceptual confusion can arise in all avenues of research, regardless of whether one follows a natural science approach or an alternative path.
We contend that conceptual issues in psychology are greatly overlooked; and, unfortunately, questions such as “Should psychology follow the methods and principles of the natural sciences?” potentially mask enduring conceptual confusions, as they place focus on principles and methods while bypassing conceptual clarity. Conceptual clarity should be antecedent to the carrying out of empirical research. Prior to determining whether a particular question calls for a scientific method of examination, the psychological concepts at hand should carefully be considered. Only after such consideration can a researcher determine whether a natural science method of knowledge acquisition is fitting for their purposes. We thus address the focal question of this special issue by examining the conceptual confusions that result from unreflectively accepting the dominant methods and principles of the natural sciences. To be clear, we are not arguing that psychological researchers necessarily are not, or should not strive to be, natural scientists; rather, we see our role as illuminating and then clarifying conceptual confusions that may arise when embarking on this pursuit. Our goal is to underscore the importance of conceptual clarity in determining whether natural science methods and principles might be sensibly pursued in specific circumstances.
Principles of natural science (as summarized by two psychologists)
We are mindful of the fact that descriptions of “natural science” vary across academic communities. We take as a starting point the view that “natural science” encompasses both the broad traditional scientific domains of physics, chemistry, and biology, as well as the specialized subdisciplines within those broad categories (e.g., botany) and interdisciplinary sciences (e.g., earth science). As scholars trained in the social science tradition, we are not entirely confident that we are in the best position to speak for natural scientists as to the specific principles they accept and follow in their work. However, as scholars trained on the model (at least presumed) of the natural sciences, and as psychologists who have thought deeply about the philosophical underpinnings of practiced research in psychology, we believe that we are in a good position to describe—at a high level—the ontological and epistemological 1 principles that psychological researchers take to be fundamental to science. Our aim here is not to provide a thorough summary of all potential natural science principles; rather, it is to provide a starting point from which to develop our arguments. We do not argue that these principles are followed by all natural scientists today. In fact, one might argue that psychologists’ perceived view of science is far more deterministic and objectivist than that of modern-day natural scientists (Gigerenzer, 1987). However, since the question around which this special issue is oriented concerns whether psychology should follow the methods and principles of the natural sciences, we summarize the natural science principles we believe to be generally presumed by psychological researchers.
A basic ontological principle of natural science is captured in the doctrine of objective realism, 2 according to which the subject matter of science pertains to real objects, processes, events, and properties thereof, the existence or natures of which are not contingent on humans’ capacity to perceive, conceptualize, and hold knowledge of them. Materialism, mechanism, determinism, and universalism are four additional ontological principles that often accompany objective realism. Briefly, materialism presupposes that “reality” is composed only of physical matter and that everything else (e.g., thoughts, emotions, will) is ultimately reducible to physical matter and relations thereof. According to mechanism, reality is generally composed of material and physical basic building blocks that function in a regular and predictable manner. Determinism is the belief that all events in the world are effects of prior events (or chains of events), that is, are inevitable consequences of antecedent states of affairs, and that these occur in a predictable (i.e., lawful) way. Universalism is the view that there exist general truths (i.e., true in all cases) about the objects of the material and physical world and their effects.
A fundamental epistemological principle of natural science is that the objects that make up the subject matter of a science are knowable through both observation (direct and mediated) and reason. Thus, any assertion about the object under study must be coherent (i.e., make sense) and be confirmable (or falsifiable) by reference to empirical evidence. These are the basic premises of epistemological objectivism. The principle also assumes that, within and across domains of study, the relevant objects are known and understood to varying degrees and that knowledge of a given object is, at any one time, more or less complete, and is always revisable in the face of compelling and stable empirical evidence. Also assumed is the view that the perceptual systems of humans—scientists notwithstanding—are limited, thus impeding the ability to observe directly or clearly (or, more generally, perceive) all the objects relevant to the subject matter under study, and “mediated” observation may be needed (e.g., microscopes, telescopes, computer generated images). In a related way, the conceptual systems of humans, although also limited, enable scientists to denote, designate, and otherwise symbolize the objects under study; theorize about the subject matter; deduce and empirically test hypotheses from theory; and generalize, extend, and apply empirical findings. Moreover, despite limitations of human perceptual or conceptual capacities, it is assumed that the methods of science mitigate to a great extent the obscuring of “things as they are” by “things as they appear” to the scientist. Two additional basic epistemological principles of natural science are that: (a) scientific theories describe, predict, and provide causal explanations for the object under study and (b) the relative worth of a scientific theory may be evaluated by a range of different criteria (including clarity, internal coherence, parsimony, etc.); however, the primary criterion for judging a theory to be a good theory is the extent to which it is compatible with what actually occurs and is observed.. As such, science is taken to be a pursuit of empirical discovery. It is the activity of uncovering the unknown.
Why conceptual clarification? Distinguishing everyday versus technical concepts
In responding to the focal question of this special issue, we aim to provide an argument for emphasis on conceptual clarity, rather than a focus on methods and principles. We take the stance that conceptual clarity comes about through conceptual analysis of relevant concepts. In this case, these would include both the psychological and methodological concepts that come into play when a psychological researcher attempts to follow perceived principles of natural science. Our approach to conceptual analysis is based on Wittgenstein’s (1953) later philosophy, while our interpretation of his work is heavily influenced by Baker and Hacker (1982, 2009; see also Bennett & Hacker, 2022). In simplest terms, conceptual analysis involves determining the contexts in which a concept, or set of concepts, has a sensible use within a community of language speakers. More specifically, it involves describing the logical space of a concept, that is, to make clear the circumstances in which it could reasonably be called upon to denote, describe, express, and so forth, something by competent users of a language (and lexicon). As such, conceptual analysis can be used to reveal the presence of confusing or ambiguous uses of terms, or philosophical muddles. It can also point in the direction of resolutions to such confusions or muddles, through the careful elucidation of the concepts involved.
An important distinction to be highlighted here is that between conceptual and empirical matters—and conceptual and empirical analysis—within a domain of inquiry. Whereas empirical matters are matters of fact (descriptions, causal explanations, and predictions) which can be investigated through the empirical methods and analyses that form the basis of scientific practice, conceptual matters require us to consider the meanings of concepts and patterns of reasoning (Baker & Hacker, 1982), including those used in scientific discourse. Unlike empirical analyses, conceptual analyses do not produce new knowledge about phenomena; rather, they help us to understand “the possibilities of phenomena” (Wittgenstein, 1953, p. 47e). In other words, conceptual analysis allows us to understand whether a given claim has a sense, that is, a use that coheres with the logical spaces of the concepts at play. Moreover, because empirical analysis requires clarity of concepts relevant to its employment, conceptual analysis is logically prior to empirical investigation (Baker & Hacker, 1982).
Another important distinction we rely upon in this article is between ordinary and technical concepts (Baker & Hacker, 1982). In simple terms, ordinary concepts are those that are common in everyday language-use (think, feel, see, lose, eat, suffer, fast, tall, anxiety, pencil, etc.). Another way to think of these is as familiar concepts that are common outside of technical discourse. Technical discourse is specific to the constrained uses of concepts within a community of language users, for example, particle physicists. There are plenty of examples of technical concepts in the natural sciences that serve the purposes of identifying, measuring, and explaining physical and material phenomena (e.g., kinetic energy, gravitational force, mass, velocity, chemical elements, etc.).
As previously described, the received ontological viewpoint of psychology is in line with the natural science principle of objective realism. As such, it is commonly presumed by psychologists that psychological phenomena precede the language used to describe, categorize, and represent them. For example, it is widely accepted practice to alter (in an attempt to refine) the definitions of ordinary psychological concepts based on the results of empirical research. Wittgenstein (1953) went to great lengths to demonstrate how this perspective cannot account for the complexity of the relationship between language and meaning. Although one cannot deny that the facts of the world place limits on sensible possibilities regarding language-use (i.e., that grammar is “responsive to reality”; Baker & Hacker, 2009, p. 339), it would be a mistake to assume that psychological concepts, or their meanings, are direct representations of such facts (i.e., that grammar is accountable to reality; Baker & Hacker, 2009). We wish to emphasize that this view does not reject the facts of reality, or claim that reality can be reduced to grammar. Rather, it simply implies that conceptual questions (e.g., “What is depression?” “What is happiness?”) are separate from empirical ones, and, importantly, they cannot be answered through empirical means. For a claim to reasonably be judged as factual, one must presuppose concepts that designate the facts in question. For example, “The results of our study showed that Danish participants are happier than American participants” presupposes the meanings of the concepts of “study” and “happiness,” among others. Again, this does not imply that one can randomly create any new meaning for a concept based on how they wish to employ its grammar, or that language constructs the facts of the world. The rules of grammar must correspond to the facts of reality; however, they are not accountable to empirical observation once they have been established.
It is beyond the scope of this article to chart a full description of Wittgenstein’s views in detail. Readers are encouraged to refer to Maraun (1998a, 1998b, 2021) and Ter Hark (1990) for further discussion on the relationship between grammar and facts. The central point posited by Wittgenstein (1953) that we adopt here is that the meanings of concepts are specified by their grammars, that is, the linguistic conventions that guide their uses in specific contexts. This point guides conceptual analysis in two ways. First, it allows one to examine methodological (technical) and psychological (ordinary) concepts by examining their sensible uses. Second, it cautions against the slippery slope of mixing technical and ordinary uses of concepts. It is true that technical concepts sometimes “spill over” into everyday language (e.g., using the term “gravity” in a metaphoric sense to describe the feeling of being emotionally impacted in a song or poem). This is no surprise, given that all members of the scientific community are also members of everyday society. However, distinguishing between technical and everyday senses of terms provides clarity about meaning and use in reminding researchers that technical and everyday senses are not, in most cases, interchangeable (e.g., technical gravity and metaphoric gravity; Ter Hark, 1990). As such, the context of use is also important to consider when determining whether a concept is being used in an everyday (including metaphorical) or technical sense. Bennett and Hacker (2022) used an asterisk (*) to denote the use of technical concepts in psychology to differentiate these from ordinary ones. For example, when a social psychologist operationally defines happiness in relation to items on a scale, we can refer to this new technical concept as happiness* to differentiate it from the ordinary concept of happiness which has a much more nuanced and complicated meaning in everyday discourse.
To readers unfamiliar with the works of Wittgenstein, it may seem oversimplified to focus heavily on words and their meanings. The reason why we emphasize these is to consider the kinds of concepts that psychologists use and whether they are akin to the kinds of concepts that are typically used by natural scientists. The intent is not to reduce psychology or science, more broadly, to language-use; rather, it is to illuminate possible ways in which it may be sensible to draw comparisons between the (primarily ordinary) concepts used in psychology and the (primarily technical) ones used in the natural sciences. In the same vein, conceptual analysis also allows us to examine whether certain methodological concepts are appropriate to the epistemological aims that psychologists who wish to pursue natural science principles strive for.
The use of psychological and methodological concepts in scientific psychology
We now turn our attention to reviewing how psychological and methodological concepts come into play when natural science principles are pursued in psychological research. Much of this work has previously been conducted in detail by scholars such as Baker and Hacker (1982), Bennett and Hacker (2022), Lamiell (2003, 2019), and Maraun (1998a, 1998b, 2021). We draw heavily from these works in making our arguments. It is important to note that our focus is on the analysis of concepts; we are not prescribing new methods or sets of principles, nor are we advocating for a specific method or set of principles. However, conceptual analysis will often make clear where certain methods and principles may serve to be problematic.
Misuse of psychological concepts in measurement practice
A science of psychology that models itself after the natural sciences will likely presume that measurement is both a possible and necessary component of research. Numerical systems are by their very nature precise and amenable to rigid rules. Representing the domain under study in numerical—or quantitative—form gives at least the appearance of objective descriptions of the subject matter. Since science is considered a pursuit of discovery, a natural science sense of measurement is taken to be the process of “discover[ing] facts about magnitudes” (Michell, 2004, p. 119); thus, the attributes that are purportedly measured through this practice must have quantitative structure to be discovered. However, asking whether psychological phenomena have quantitative structure to be discovered is a nonsense question (Tafreshi, 2022). As was argued by Wittgenstein (1953), psychology is largely concerned with ordinary concepts, rather than technical ones. The grammatical rules that clarify the sensible uses of these concepts are not embedded in measurement practices, nor are their meanings contingent upon empirical observations. As such, there is no sense in asking whether psychological attributes are quantitative, or have quantitative structure awaiting to be discovered. Detailed analyses of a Wittgensteinian approach to examining the use of measurement in psychology have been previously given by Maraun (1998a) and Bennett and Hacker (2022). Recently, Franz (2022) drew heavily on Bennett and Hacker (2022) in reviving this perspective to examine current debates in the measurement literature. In our view, the slippage between ordinary and technical concepts is the critical point that summarizes why speaking of psychological phenomena in measurement terms amounts to nonsense.
Attempts to apply measurement in psychology always require the creation (often through operationalization practices) of new technical concepts (e.g., psychopathy = PCL-R score, Hare, 2003; latent memory = time duration between presentation and recall). Such operationalization is not necessarily a problem. Rather, the assumption that operationalizations are interchangeable with the initial concepts from which they were derived is where the confusion sets in (Baker & Hacker, 1982; Bennett & Hacker, 2022). The path of measurement creates confusion for psychology because it often leads to muddling of ordinary and technical (operationalized) senses of psychological terms. For example, social psychologists typically aim to study happiness in its ordinary and complicated senses. When happiness* is operationally defined as a technical concept (e.g., subjective happiness scale score; Lyubomirsky & Lepper, 1999), it is no longer the everyday sense of happiness. It may be derived from the everyday sense; however, it would be both ontologically and methodologically reductive to assume that happiness* can be used interchangeably with happiness (see Bennett & Hacker, 2022).
Moreover, psychological researchers who attempt to employ measurement by means of numerical representation are not “discovering” the magnitudes of psychological traits (processes, etc.); any such “discovery” (if we were to afford it that) would be about the new technical operationalization of the initial concept of interest. For example, one might, after correlating total scores on a measure of happiness* with those on a measure of well-being*, find that they positively correlate. The empirical occurrence here is the positive correlation of total scores on two measures based on the operationalizations: happiness* and well-being*. It is not a discovery about the logical relations between the ordinary senses of happiness and well-being, or of the natures of these phenomena. Based on the rules set out by the grammar of everyday concepts, we already know that it makes sense to speak of happiness and well-being as closely related concepts. Thus, any potential claim of an empirical realization of the meaning of happiness or its relation to the meaning of another ordinary concept would simply amount to tautology (see also Gergen, 2018; Smedlsund, 2016).
It might be argued that although operationalizations cannot capture the complexity of a psychological concept, they can, nonetheless, be considered one useful extension of an everyday concept. Indeed, it is sensible to speak of happiness in terms of degrees. For example, someone might proclaim, “I am much happier today than I was yesterday.” Similarly, a clinician might even use a rating scale to determine that their client’s level of happiness is quite low given their circumstances. We do not deny that these are sensible uses. The confusion is not due to whether it is sensible to speak of psychological phenomena in terms of ordinal scales. Rather, the confusion occurs when one slips between the constrained possible usages of a technical sense of happiness*, which are required to administer such scales, and the more complex everyday senses of happiness (Bennett & Hacker, 2022; Maraun, 1998b). We presume that psychologists are ultimately interested in examining the latter. Moreover, the constrained technical senses are typically parasitic upon the ordinary senses and, thus, their meanings are presupposed in any such slippages.
Another common argument is that our “folk” (everyday) concepts are insufficient for capturing the true nature of psychological phenomena and that, with time, researchers will refine current concepts to be more amenable to measurement practices. For example, Michell (2022) posited that “it is possible that our current conceptualisation misfits the phenomena of anxiety in all its richness, in which case our empirical investigations might alter our concept, including its logical relation to quantitative structure” (pp. 153–154). Michell (2022) further compared psychological concepts with the technical concept of temperature, arguing that psychological concepts may undergo the same conceptual change that advanced the notion of temperature in science. Here, Michell (2022) confuses ordinary and technical concepts, implying that if the perceived shortcomings of ordinary psychological concepts are refined into precise technical concepts, then measurement of psychological phenomena may be possible. While it is true that the rules of grammar can and often do change over time, the notion that our current conception of anxiety is a “misfit” presumes that the meaning of anxiety exists independent of language use. However, the very fact that Michell (2022) and other psychologists can discuss the notion of “anxiety” and potentially create items for an anxiety scale presupposes that anxiety has a meaning (and that researchers are already aware of what it is). One must already know how to sensibly use the term anxiety to create an anxiety scale. There is no “discovery” to be made about anxiety that would refine our understanding of it. Any changes to the grammatical rules pertaining to sensible uses of the term anxiety in everyday discourse would change the conceptual foundations upon which it is based, thereby changing the concept itself (Baker & Hacker, 2009). The everyday senses of anxiety are not akin to the technical senses of temperature, which are defined in terms of measurement practices (e.g., the Kelvin scale). Perhaps a different, technical (operationalized) sense, of anxiety* would be more comparable; however, to slip between the technical sense of anxiety* and anxiety in its everyday senses would lead to conceptual confusion, not conceptual refinement. Unlike temperature on the Kelvin scale, anxiety is not a term embedded in measurement practices; its meaning is not derived from empirical investigations but rather from grammatical rules (often tacit but embedded in norms of use; Maraun, 1998a). Moreover, the promissory note that folk psychological concepts are merely prescientific descriptions of the “external” (i.e., material, physical) world (including human brains and bodies, and cognitive “mechanisms” and “systems”) ignores a fundamental feature of ordinary psychological concepts, namely, that they are constitutive of the phenomena they are used to describe; to reduce or eliminate them would strip scientific psychology of its subject matter (Hacker, 1996).
Misuse of technical concepts in experimentation and statistical inference
In the previous section, we discussed how equating ordinary and technical senses of psychological concepts through operationalization can lead to conceptual confusion. Here, we maintain our core argument, centering on problems that arise when psychologists slip between uses of everyday and technical concepts. However, we shift our focus to the misuse of technical, rather than everyday, concepts when employing experimentation in conjunction with statistical inference. We highlight the misuse of aggregate-level statistical concepts to draw individual-level conclusions as one example of concept misuse. Given that Lamiell (2003, 2019) has written extensively on this topic, we draw primarily from his works in making our arguments.
It is common knowledge that the gold standard of mainstream psychological research is the experimental method coupled with statistical inference. Psychologists who aim to mimic the natural sciences employ statistical inference as if it is a requirement of all research practice (Lamiell, 2013, referred to this tendency as “statisticism”). Statistical concepts are technical in nature—they are rigidly defined, and their meanings are derived from their technical (often mathematical) uses. Some statistical concepts, such as the “average,” spill over into everyday language. However, this does not alter their technical meanings when used in the context of statistical practice. Unfortunately, in interpreting results from statistical inferences, psychologists, in their pursuit of the natural science principles of determinism and universalism, often misuse technical statistical concepts. For example, the meaning of the average value, or the arithmetic mean, is specified by its technical use. It is the sum of all numerical values of interest divided by the total number of values. Therefore, it is a statistic that is computed based on the aggregation of numerical values. In the language of statistics, there is no wiggle room for interpreting the mean. Yet, in interpreting the results of statistical inferences made based upon the mean, psychologists overwhelmingly attempt to draw deterministic (i.e., lawlike) and universal conclusions that are applicable to all individuals (Lamiell, 2003, 2019). For example, it is not at all uncommon to find in published psychological research a conclusion on the basis of a positive correlation (an aggregate function) between scores on measures of two putatively related psychological variables, X and Y, that individuals high in X will be high in Y, but such inferences are not warranted by aggregate analysis, only that, the probability of being high on Y increases with higher scores on X. And, certainly, no conclusions can be made about the probability of this event for a single individual or the causal relationship between X and Y based on estimated correlations between measurements of X and Y.
Lamiell (2003, 2019) has argued extensively that the correct way to interpret the mean would be to put forth generalizations that are true on average. Inferences based on aggregated statistics do not allow psychologists to assume the principles of either determinism or universalism. To make claims that are generalizable to all cases (or universal), inferences would need to be based on statistics computed on individual data and would require experimental designs that warrant causal inferences. Lamiell (2003, 2019), following Bakan (1967), has detailed this clarification between aggregate-type (true on average) and general-type (true of all cases) propositions. Further, Lamiell (2003) described how, historically, some psychologists (e.g., Wundt) who have aimed to make general-type claims have turned to N = 1 research. As such, one might wonder if N = 1 research can bring psychology closer to the scientific principles of determinism and universalism. Indeed, there has been a resurgence of interest in N = 1 research and within-person design over the past couple of decades, which, in our view, is a more adequate method of achieving the aim of general-type propositions (Hamaker, 2012; Molenaar, 2004). However, the Wundtian model of N = 1 research was abandoned by psychologists in the early 20th century for population-level research (à la Galton, 1892) partly because of the conflicting priorities of psychological researchers (Danziger, 1990; Lamiell, 2003). While psychologists wish to make universal claims, they also tend to uphold determinism as a principle of psychological science. That is, they wish to make universal claims, but they also seem to want to provide cause–effect explanations. However, it is difficult (if not nearly impossible) to draw conclusions that are both universal and causal in nature about psychologists’ preferred unit of observation: persons. This is due to both the richness/complexity of psychological phenomena as well as the constraints of measurement and statistical approaches. For example, one cannot carry out a randomized controlled trial with an N of 1. Nonetheless, in attempting to achieve these scientific principles, psychologists have adopted a neo-Galtonian approach to research that combines aggregate-level statistics with the experimental method (Danziger, 1987, 1990). Unfortunately, this maneuver has led to conceptual confusion concerning aggregate-level statistical concepts, such as the mean and correlation coefficient, now frequently being misinterpreted as individual-level concepts (Lamiell, 2003; Tafreshi, 2021).
The interpretation of technical statistical concepts using everyday senses is not only restricted to means and correlations. The concept of “significance” is another potent example. The technical (statistical) sense is defined as the long-run probability of obtaining a given result of a statistical test (i.e., the “observed” test statistic) that is below a prespecified threshold (i.e., “alpha”), conditional on the null hypothesis being true. Clearly, this statistical sense of “significance” does not have the same meaning as the everyday sense of the term, the latter of which connotes importance, relevance, noteworthiness, and the like. Yet, achieving a statistically significant result appears to be a goal post in psychology, which signals a conflation of technical and ordinary senses. The claim that a result of a “statistically significant” test of a statistical hypothesis is a significant finding in the ordinary sense will require much more justification than simply deeming the result to be “significant” (i.e., an important scientific discovery). There is a long history of detailing the misuses of “significance” within psychological science discourse (e.g., Lykken, 1968; Meehl, 1978) and it remains a criticism in current debates on the need for statistical reform in psychology (e.g., Perezgonzalez, 2015; Simmons et al., 2011). The primary takeaway for us is that statistical concepts are often being “stretched” in psychological discourse to meet principles of natural science (e.g., determinism and universalism) and to demonstrate that psychologists are participating in the activities of science (i.e., objective empirical discovery of noncontingent objects, events, and properties thereof). However, the permissible usages of these technical concepts often do not (and cannot) sensibly include these overextensions.
Making sense of natural science principles and methods in psychological research
We previously described the major principles that, we believe, psychological researchers who follow the natural science model of research aim to pursue. We demonstrated how the desire to pursue principles such as objectivism (both ontological and epistemological), determinism, and universalism can often lead to research practices wherein methods are prioritized, and conceptual questions are overlooked. Said another way, most psychological researchers simply presume that the psychological domain can be approached with the same methods used by natural scientists, just applied to the subject matter of psychology. A consequence of this has been the misuse of concepts (both substantive and methodological) due to the equivocation of ordinary and technical senses. The solution, in these cases, is not to consider alternative principles or methodological approaches. Rather, it is to take a step back and examine the concepts that are of relevance in a particular scenario. For instance, does it make sense, based on the logic of our ordinary grammar, to think of psychological traits as mental versions of physical processes and mechanisms? Once clarity around relevant psychological concepts is gained, decisions regarding whether it makes sense to approach one’s initial question(s) through empirical methods (and what those methods ought to look like) can also be made. Subsequently, any relevant methodological concepts can also be subject to conceptual scrutiny.
Although tempting, it can be misleading to prioritize principles of science, such as ontological assumptions pertaining to the existences or natures of phenomena under study. For example, if a social psychologist interested in studying “happiness” prioritizes ontology, determining that, based on the principles of objective realism, happiness can be conceived of as a real process with properties that are reducible to physical matter, they might attempt to measure happiness as a property of the brain. This may appear reasonable to do, based on the ontological position that they have assumed. However, this emphasis on ontology would overlook important conceptual questions regarding the meaning of happiness and whether it makes sense to use the concept in a way in which it is ascribable to the brain or other physical systems (Bennett & Hacker, 2022). In other words, although the happiness researcher’s methods match their ontological perspective, their ontological perspective is based on a faulty conceptualization of the ordinary concept of happiness and the contexts it can legitimately be extended to, thus resulting in conceptual confusion. And, only through conceptual clarification can such confusions be resolved. No amount of additional empirical work, or implementation of novel empirical methods, will sort out whether the principles that are upheld in a given domain of inquiry are reasonable or not. However, once the concepts at hand have been clarified, the relevant empirical questions and the methodological options (including limitations thereof) for pursuing them will also become clear.
To avoid prescribing a “one size fits all” approach and thus falling into dogmatic practices, we advocate individual researchers conducting analyses of the concepts relevant to their specific questions. We also described how both psychological and methodological concepts can potentially be misused when natural science principles are prioritized. Given that psychological concepts would need to be clarified prior to the engagement of scientific practice, such concepts would logically require examination prior to the consideration of methodological concepts.
Concluding remarks
Given our review, one might presume that we would conclude that psychology cannot coherently follow the natural sciences. However, as previously mentioned, our aim has been to focus on conceptual clarity, rather than the prescription of empirical approaches. Although our review has demonstrated that following natural science principles and methods is often not a good “fit” when the domain of study is represented largely by irreducible ordinary psychological language, it is not an exhaustive review. Rather, it is meant to provide examples of how and why conceptual clarity is important to consider prior to making decisions about research processes. There is always the possibility that a psychological researcher is interested in a question that does coherently lend itself to natural science empirical approaches. For example, empirical methods used within the neuroscientific domain examining the structures and functions of the brain can reveal why some humans are incapable of experiencing normal human emotion (Bennett & Hacker, 2022). However, any researcher examining such a question would first need to clarify meaningful uses of the concept “emotion,” to ensure that they are using it in sensible ways within their empirical pursuits. More generally, we recommend that individual researchers conduct analyses of the concepts relevant to their particular research scenarios, and, ultimately, make reasoned—rather than dogmatic—decisions regarding research practice.
Finally, in advocating for the primacy of conceptual analysis in psychology, we wish to be clear that we are not arguing that psychological inquiry is nothing more than conceptual analysis. Certainly, there are empirical questions in psychology that cannot be addressed through the examination of concepts. Conceptual clarification has no bearing on what might actually occur. For example, we might correctly apply the grammatical rules pertaining to the use of anxiety-words when judging someone as being anxious based on their worried look, shallow breathing, and wringing of hands, but it may later turn out that this individual was participating in role-play and that these behaviors were not expressions of their truly being anxious. Conceptual analysis can reveal what it means to be anxious but not that someone acting in accordance with these grammatical criteria is, in fact, anxious. It can help clarify possibilities with regard to language-use, but it cannot explicate what will occur in the world. Nonetheless, the primary point we wish to reiterate is that the question “Should psychology follow the methods and principles of the natural sciences” should be answered on a case-by-case basis and only after relevant conceptual concerns have been addressed.
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
