Can psychology become a science?

I am profoundly grateful to Tom Bouchard for helping me learn to think scientifically. Scientific thinking, which is characterized by a set of safeguards against confirmation bias, does not come naturally to the human species, as the relatively recent appearance of science in history attests. Even today, scientific thinking is in woefully short supply in many domains of psychology, including clinical psychology and cognate disciplines. I survey five key threats to scientific psychology – (a) political correctness, (b) radical environmentalism, (c) the resurrection of ‘‘common sense” and intuition as arbiters of scientific truth, (d) postmodernism, and (e) pseudoscience – and conclude that these threats must be confronted directly by psychological science. I propose a set of educational and institutional reforms that should place psychology on firmer scientific footing. 2010 Elsevier Ltd. All rights reserved. 1. Can psychology become a science? When I entered graduate school in psychology at the University of Minnesota in the Fall of 1982, I was a bright-eyed, bushy-tailed 21 year-old eager to learn about the mysteries of the mind. I was brimming with energy, intellectually curious, and deeply in love with psychology. Yet despite my undergraduate education at a superb institution, Cornell University, something important was conspicuously absent from my intellectual repertoire, although I did not realize it at the time. I had not learned how to think. As one symptom of my dysrationalia, to use Stanovich’s (2009) term, I confidently held a host of profoundly misguided beliefs about individual differences. Among other things, I was certain that: Genetic influences on most psychological traits are trivial. Genes and environments always interact. Genes and environments cannot be separated. IQ tests are invalid for predicting cognitive performance. IQ tests are strongly biased against minorities. At the time, it never occurred to me that some of these beliefs were not only poorly supported, but contradictory. For example, it never crossed my mind that if one cannot separate the influences of genes and environments, there is no way of ascertaining whether genes and environments interact statistically. Nor did it cross my mind that for IQ tests to be biased against certain subll rights reserved. groups, they would need to possess above zero validity for at least one subgroup. Of course, a naïve graduate student can perhaps be forgiven for such logical errors, especially one embarking on his training nearly three decades ago. Yet as Faulkner (1951) noted, ‘‘The past is never dead. In fact, it’s not even past.” Even today, in the pages of our journals and newsletters, we can find similar misunderstandings of individual differences psychology. Witness, for example, two recent passages from the pages of the APS Observer, the newsletter of the Association for Psychological Science: ‘‘. . .partitioning the determinants of behavioral characteristics into separate genetic versus environmental causes is no more sensible than asking which areas of a rectangle are mostly due to length and which to width” (Mischel, 2005, p. 3). ‘‘. . .this approach [traditional behavior genetics] does not escape the nature–nurture dichotomy, and it perpetuates the idea that genetic and environmental factors can be accurately quantified and their relative influence on human development measured. . .genes and environment are always interacting, and it would be impossible to consider one without the other” (Champaigne, 2009, p. 2). Both quotations confuse the transaction between genes and environment within individuals with the separate influences of genes and environment across individuals (Rowe, 1987). Mischel’s assertion, like many others in the literature (e.g., Ferris, 1996; LeDoux, 1998), implies erroneously that one cannot examine the question of whether good quarterbacks are more important to a football team’s success than are good receivers, because quarterbacks ‘‘depend on” receivers to function, and vice versa. Yet it is 282 S.O. Lilienfeld / Personality and Individual Differences 49 (2010) 281–288 entirely possible to partition sources of variance across individuals even when these sources ‘‘depend on” each other within individuals (Waldman, 2007). Chaimpaigne’s claim exemplifies the same error, and compounds it by asserting simultaneously that (a) genes and environments always interact, but that (b) one cannot separate or quantify the relative influences of genes and environments, despite the fact that one cannot ascertain whether genes and environments interact statistically without separating them as sources of variance. Incidentally, I strongly suspect that as a beginning graduate student, I would have found both of the aforementioned quotations persuasive, in part because they dovetailed with my own biases against genetic influences, or at least genetic main effects, on behavior. It was not until my second year of graduate school at Minnesota, when I enrolled in Tom Bouchard’s course on individual differences, that I first began to learn to think scientifically – that is, to try to put aside my biases in an effort to align my beliefs more closely with reality (in this respect, I am an unabashed adherent of the correspondence theory of truth; O’Connor, 1975). Tom taught me that political correctness has no place in science: The desire to discover the truth must trump the desire to feel comfortable (see also Sagan, 1995). Tom also taught me that we must be courageous in facing up to evidence, regardless of where it leads us, and that as scientists we must prepare to have our preconceptions challenged, even shattered. More than anything, Tom inculcated in me a profound appreciation for intellectual honesty, which B.F. Skinner (1953) regarded as the ‘‘opposite of wishful thinking” (p. 12). For this wisdom, which I have always tried to take to heart as a researcher and teacher, I will forever be grateful. 2. The unnatural nature of scientific thinking Why did I begin this article by presenting misguided statements by myself and other psychologists? To make a straightforward point: Scientific thinking does not come naturally to any of us. In many respects, science is ‘‘uncommon sense,” because it requires us to set aside our gut hunches and intuitions in lieu of convincing data (Cromer, 1993; McCauley, 2000; Wolpert, 1993). Even many great thinkers have failed to grasp this profound truth. Huxley (1902), Darwin’s ‘‘bulldog,” wrote that ‘‘science is nothing but trained and organized common sense” and mathematician-philosopher Whitehead (1916) wrote that ‘‘science is rooted in the whole apparatus of commonsense thought.” In contrast, other scholars, including eminent psychologists, have offered a diametrically opposed perspective, one more consonant with that I present here. Titchener (1929) maintained that ‘‘common sense is the very antipodes of science,” and Skinner (1971) asked rhetorically, ‘‘What, after all, have we to show for non-scientific or prescientific good judgment, or common sense, or the insights gained from personal experience? (p. 160).” Skinner’s characteristically blunt answer: ‘‘It is science or nothing” (p. 160). As Cromer (1993) noted, ‘‘All non-scientific systems of thought accept intuition, or personal insight, as a valid source of ultimate knowledge. . .Science, on the other hand, is the rejection of this belief, and its replacement with the idea that knowledge of the external world can come only from objective investigation (p. 21).” Cromer’s insightful observation helps to explain why science is a relatively recent development in history. Science requires us to override more automatic, effortless, and intuitive modes of thinking with more controlled, effortful, and reflective modes of thinking (Stanovich, 2009). According to many scholars, science arose only once in world history, namely in ancient Greece, reappearing in full-fledged form in the European enlightenment (Wolpert, 1993). Even the concept of control groups, which we take for granted today, did not emerge in psychology until the early 20th century (Dehue, 2000). The necessity of control groups is decidedly unintuitive, as these groups are designed to eliminate alternative explanations that lie outside of our immediate sensory awareness. Our commonsense realism or ‘‘naïve realism” – the seductive but erroneous belief that the world is exactly as we see it (Ross & Ward, 1996) – tells us that if a group of depressed clients improves following therapy, we can conclude that the therapy worked. Our naïve realism assures us that ‘‘we have seen the change with our own eyes” and that ‘‘seeing is believing.” Yet these conclusions are erroneous, because they do not control for a host of rival explanations that lurk in the causal background, such as regression to the mean, placebo effects, spontaneous remission, effort justification, and the like (Lilienfeld, Lohr, & Olatunji, 2008). 3. What is science, anyway? Up to this point, I have said little or nothing about what science is. Some scholars insist that any attempt to define science is doomed to fail, as the specific methodological procedures used in one domain (e.g., astronomy) often bear little or no superficial resemblance to the procedures used in others (e.g., psychology; Bauer, 1992). Yet this argument overlooks the possibility that certain higher-order epistemic commonalities cut across most or all scientific domains. I side with several authors who maintain that science is a set of systematic safeguards against confirmation bias, that is, the tendency to seek out evidence consistent with our hypotheses and to deny, dismiss, or distort evidence that runs counter to them (Hart et al., 2009; Nickerson, 1998; see also Lilienfeld, Ammirati, & Landfield, 2009). Nobel-prize winning physicist Feynman’s (1985) aphorism that the essence of science is ‘‘bending over backwards to prove ourselves wrong” succinctly embodies this view, as does Skinner’s (1953) conclusion that science mandates a ‘‘willingness to accept facts even when they are opposed to wishes” (p. 12). This emphasis on disconfirmation rather than confirmation accords with Popperian and neo-Popperian views of the philosophy of science (Meehl, 1978), which underscore the need to subject our most cherished hypotheses to the risk of falsification. More broadly, this emphasis dovetails with the point that science is a prescription for humility (McFall, 1996) and a method of ‘‘arrogance control” (Tavris & Aronson, 2007). The adoption of scientific procedures, such as control groups, is an explicit acknowledgement that our beliefs could be wrong (Sagan, 1995), as these procedures are designed to protect us from fooling ourselves. As we all know, scientists are hardly immune from confirmation bias (see Kelley and Blashfield (2009), for a striking illustration in the domain of sex differences research). Mahoney (1977) asked 75 journal reviewers who held strong behavioral orientations to evaluate simulated manuscripts that featured identical research designs but different results. In half of the cases, the results were consistent with traditional behavioral views (reinforcement strengthened motivation), whereas in the other half of the cases, the results were inconsistent with traditional behavioral views (reinforcement weakened motivation). Even though the Introduction and Method sections of the articles were identical, Mahoney found that reviewers were much more likely to evaluate the study positively if it confirmed their views (quotations from the reviewers included ‘‘A very fine study” and ‘‘An excellent paper”) than disconfirmed them (quotations from the reviewers included ‘‘A serious, mistaken paper” and ‘‘There are so many problems with this paper, it is difficult to knowwhere to begin”). Still, because scientific methods themselves minimize the risk of confirmation bias, the inevitable shortcomings of the peer review process (e.g., Peters S.O. Lilienfeld / Personality and Individual Differences 49 (2010) 281–288 283 & Ceci, 1982) tend to be corrected over time by the force of consistently replicated findings (Lykken, 1968). 4. The troubling state of science in clinical psychology and