Chapter 3 – Nevermind
Reading 3.6 – Why Behavior Analysis is a Natural Science1
Behavior analysts have always viewed our field as a natural science, if only because this affiliation helps differentiate it from the social sciences – an association we usually find embarrassing. B. F. Skinner’s position that behavior analysis was part of biology is consistent with our self-nomination to this esteemed club (Skinner 1978, Chapter 6). We may be happy to bask in the glow of such honored senior sciences as physics and chemistry, but what does it mean to be a natural science? By what standard do we qualify?
How do you tell them apart?
The distinction between the natural and social sciences has long been a matter of debate, and consensus is unlikely to be reached. It is certainly not as easy as proposing that the social sciences study human affairs at the level of the intact organism and the natural sciences study everything else. Were this the rule, behavior analysis would be squarely in the social sciences, and that certainly doesn’t feel right.
One alternative is to focus on methodological practices that might allow for consistent distinctions. The problem with this approach is that there are considerable variations in research methods across all sciences. Almost any practice more or less characteristic of the social sciences can probably be found somewhere in the natural sciences.
Another possibility lies in how sciences define their subject matters. The natural sciences are consistent in attending only to natural or physical phenomena – events that are known or at least strongly suspected on the basis of good evidence to actually exist. Furthermore, they attempt to explain such phenomena only in terms of other physical phenomena.
The social sciences also study a natural phenomenon – behavior – but this characterization is more than a little bit misleading. Social scientists usually study behavior only as a means to a very different end. Behavior is typically viewed as an epiphenomenon caused by mental events – the real subject matter of interest. Measures of behavior are assumed to represent these mental activities, which cannot be directly measured. It takes only passing familiarity with psychological and other social science literatures to appreciate that the theoretical underpinning of most social sciences research depends on a universe of mental phenomena whose invention is clearly cultural (Skinner, 1971; 1978, Chapter 8).
Living with a non-physical subject matter
Does the focus of the social sciences on a non-physical subject matter justify a categorical distinction between natural and social sciences? Are the effects of this difference so important as to require a formal bifurcation of scientific activities? After all, natural scientists routinely theorize about and study phenomena that cannot be directly measured, sometimes without any certainly that they actually exist. Furthermore, natural scientists view their own scientific behavior no less mentalistically than do their social science colleagues. Nevertheless, the subject matters of the natural sciences and their explanations remain fully within the physical realm. Proposing an extra-physical domain as a source of both subject matter and explanation is unique to the social sciences.
The crux of the matter is not about the use of theoretical constructs or indirect measurement, however, or even the ability of investigators to avoid mentalism in conceptualizing their own behavior. It is about the pervasive and serious consequences of pursuing a non-physical subject matter. For example, in the natural sciences theory is generally built upon and constrained by pertinent facts, laws, and empirical generalizations. Because these sciences have amassed a vast body of evidence about the way the world works, these limitations are considerable. However, in the social sciences the dominant focus on mental “phenomena” provides an endless source of speculations concerning behavior that ultimately cannot be falsified. Suh speculations, offered as theory, are more likely to ignore established evidence in favor of the predilections of the theorist, as if the theories are more important than the facts.
The impact of a nonphysical subject matter on research methods and experimental style accumulates quickly. Not surprisingly, the focus on mental “events” often results in serious problems in behavioral measurement. Furthermore, an approach to experimental design driven by inferential statistical models has evolved that is more notable for its formality than its functionality. The collective result is an experimental and analytical style that falls short of building a literature that identifies reliable relationships between behavior and environmental variables, which isn’t really the objective anyway.
Over the years, the natural sciences have spawned technologies that have dramatically transformed daily life. The social sciences? Not so much. Nor is there reason to hope that this gap will be closed. The contrasts embedded in the general style of natural and social sciences is notable. For example, natural scientists seem to take experiments as genuine efforts to discover new things about nature. Lacking a comparable history of success, social scientists may be no less serious but seem to approach research more as a means to a theoretical end than as a mechanism of discovery of natural processes. Natural science researchers are always seeking improved experimental methods, and control is the holy grail of this search. In the social sciences, statistical “control” over the data substitutes for direct control over independent and extraneous variables. In the natural sciences, theories are killed on the experimental battlefield, but in the social sciences, they die of experimental neglect and old age.
Rebels at heart
Behavior analysis is a natural science in that it scrupulously avoids the insidious temptations of mentalism in the definition and explanation of its subject matter. More generally, it also seems to function more like a natural than a social science. Experimentation leads and dominates theorizing, measurement usually respects the characteristics of behavior as a subject matter, experimental methods enhance opportunities for control, and experimental design accommodates a primary interest in discovery of natural processes.
These features are a real accomplishment, given that behavior analysis has suffered from growing up in psychology, a home where it did not belong. We’ve always been rebels, however, leaving home as soon as we could to make our own way among other natural sciences and their technologies. It feels good to be a member of this club, but we must not forget the demanding membership criteria.
Johnston, J. M. & Pennypacker, H. S. (1993). Readings for Strategies and tactics of behavioral research, second edition. Hillsdale, NJ: Lawrence Erlbaum Associates.
Skinner, B. F. (1971). Beyond freedom and dignity. New Work: Knopf.
Skinner, B. F. (1978). Reflections on behaviorism and society. Englewood Cliffs, NJ: Prentice-Hall.
1 This reading is based on Reading 1 in Readings for Strategies and tactics of behavioral research, second edition (Johnston & Pennypacker, 1993).