This article originally appeared in The Skeptic, Volume 22, Issue 2, from 2012.
The discipline of parapsychology, which was earlier known as psychical research, lasted approximately 160 years. It no longer exists. This claim will likely surprise those who call themselves “parapsychologists” as well as those readers who know that annual meetings still occur and parapsychology journals are still published. So the title of this article needs an explanation.
Parapsychology was a discipline that openly claimed to be a science and actively sought evidence for psi that would meet strict scientific criteria. Henry Sidgwick (1882), in his Presidential Address to the first meeting of the Society for Psychical Research (SPR), declared that the evidence that had been collected during the preceding 30 years demonstrated scientifically the reality of supernatural phenomena. The problem was that the majority of the scientific community did not agree. The SPR’s objective was to collect even more of the same kind of evidence until the scientific community accepted it.
The goals of parapsychology as a discipline were clear: Parapsychology wanted to gain the recognition of the scientific community as a legitimate field of science. In addition, parapsychology’s mission was to collect evidence for the existence of psi that would meet the strictest scientific criteria. These goals remained central to parapsychology’s aims until fairly recently.
During the past decade, a growing number of parapsychologists have acknowledged that these goals are unrealistic and, indeed, unachievable (Bierman, 2001; Jahn & Dunne, 2008; Kennedy, 2001, 2003; Ludacou, 2001). These neoparapsychologists, as I shall call them, acknowledge that the evidence for psi is inconsistent, elusive, and fails to meet accepted scientific criteria. Such misgivings had been expressed by many earlier parapsychologists. What make the current admissions different is that the neoparapsychologists claim that the evidence for psi in principle cannot meet scientific standards. Indeed, they argue that the inability to be captured by scientific methodology is an intrinsic property of psi.
Bierman, Jahn, Kennedy, Ludacou and those parapsychologists who both agree and disagree with them will continue to do research and write about parapsychology. But a discipline that relies on evidence that cannot meet scientific standards is no longer the parapsychology that the founders had hoped to create. It is in this sense that parapsychology, as it was conceived by its founders and promoted by its practitioners, has ceased to exist.
What About Contemporary Claims That Psi has been Proven?
Although the neoparapsychologists argue that the evidence for psi cannot be scientifically demonstrated, others, in stark contrast, insist that psi has been conclusively proven with scientifically impeccable evidence. (As far as I can tell, the advocates of these opposing claims do not acknowledge each other’s existence.) The statistician and parapsychologist, Jessica Utts (1995, p. 289), wrote that, “[u]sing the standards applied to any other area of science, it is concluded that psychic functioning has been well established.” Perhaps the most outspoken advocate of the reality of psi is Dean Radin (1997). He has maintained that “we are forced to conclude that when psi research is judged by the same standards as any other scientific discipline, then the results are as consistent as those observed in the hardest of the hard sciences!” (Radin, 1997, p. 58; italics in the original).
Despite the boldness of these assertions, their support is illusory. Both Utts and Radin rely completely on the results of meta-analyses. I conducted the first meta-analysis on parapsychological data in my critique of the ganzfeld research (Hyman, 1985). Honorton (1985) did the second meta-analysis in response to mine. The fact that we came to very different conclusions doing meta-analyses on the identical data base should alert us to the fact that meta-analyses lack statistical robustness. In spite of the fact that parapsychologists subsequently began relying on meta-analysis as the main technique to support their claims, the procedures are fraught with so many blemishes that the support they offer is vacuous.
Space limits preclude discussing all the limitations of the meta-analyses, but some of the main issues are statistical, such as the heterogeneity of effect sizes that plagues almost all of these meta-analyses. Among other things this makes the use of the Stouffer Z inappropriate and overly optimistic. It also raises questions about the meaningfulness of the combined effect size. Meta-analyses are also retrospective. Using them to argue that the evidence is replicable is misleading. Replicability means that we can use current outcomes to predict new outcomes. This is something that parapsychology has been unable to do. Rather than list several more limitations of using meta-analysis to establish that psi exists and that the supporting evidence is replicable, I will highlight just one very serious problem.
Several meta-analyses have been conducted on the ganzfeld experiments. More than 100 such experiments have been reported beginning in 1974 to the present. With one controversial exception, the various meta-analyses report combined effect sizes that are positive and significantly different from zero. An effect size is simply a difference between an actual outcome and an outcome expected by chance. This difference is “standardized” by dividing it by the standard deviation (which is a measure of the variability among the subjects in an experiment). The use of effect sizes to compare the results from different experiments makes sense if the various outcomes are conceptually coherent. Just about every meta-analysis in parapsychology violates this requirement.
Let me give an example. The autoganzfeld experiments were touted as a successful replication of the original ganzfeld experiments (Bem & Honorton, 1994). As I have previously pointed out (Hyman, 1994), the autoganzfeld experiments failed to replicate the original experiments. The original ganzfeld experiments all used static targets. The meta-analysis of these experiments yielded a highly significant combined effect size. The autoganzfeld experiments used both static and dynamic targets. The resulting effect size (combining both types of targets) was approximately the same size as that for the original ganzfeld experiments as well as statistically significant. This is the basis for declaring a successful replication. The static targets, however, when considered separately, had an effect size
consistent with zero. In addition, the outcome for these static targets, which are the most relevant to the original experiments, was significantly different from the outcome for the original experiments. In other words, the autoganzfeld failed to replicate the original experiments. Bem and other parapsychologists simply ignored my argument and continued to claim that the autoganzfeld experiments were a successful replication. The neoparapsychologists agree with my position on this issue.
Broughton and Alexander (1997) attempted a direct replication of the autoganzfeld experiments. They even used some of the same equipment that Honorton and his colleagues had employed in their experiments. The result was a failure to replicate not only the overall results of the first autoganzfeld experiments, but also all of the secondary effects that were originally reported. The Broughton and Alexander experiment failed to replicate both the autoganzfeld experiments as well as the original ganzfeld experiments. Despite this, all of the major meta-analyses of the ganzfeld experiments include the original ganzfeld experiments, the autoganzfeld experiments and the Broughton and Alexander experiments. When you include the effect sizes from these three data bases, the combined effect size is both positive and “significant”. This is interpreted by Radin, Utts and others as proof that psi exists and that the supporting evidence is replicable.
An unfortunate consequence of relying upon meta-analyses to support claims of psi is that experiments that blatantly fail to replicate one another can be combined in a way that seems to say otherwise. Ironically, the very same meta-analyses that are used by some parapsychologists to boldly proclaim that the evidence for psi is replicable have been used as one of the major supports for just the opposite conclusion (Bierman, 2001). Bierman considered many of the major meta-analyses that have been done in parapsychology. For each meta-analysis he fitted regression lines to the effect sizes plotted against the year they were obtained. In all cases the slopes of the regression lines were negative. In other words, they all demonstrated a decline effect. At the beginning of each programme of parapsychological research, the effect sizes tended to be positive and significant, but as the years progressed, the effect sizes in each research programme tended to become zero. This “erosion of evidence” as some parapsychologists call it, was also noticed by many parapsychologists before its confirmation by the meta-analyses.
Some of the neoparapsychologists realize that if psi were a real effect, at least in the sense understood by orthodox science, that instead of a decline, effect sizes should show an incline over time.
Is the Elusive Nature of the Evidence an Inherent Property of Psi?
As I have indicated, several contemporary parapsychologists have marshalled impressive amounts of evidence to demonstrate that the evidence for psi is non-replicable, inconsistent, and elusive (Bierman, 2001; Jahn & Dunne, 2008; Kennedy, 2001, 2003; Ludacou, 2001). Kennedy (2003) entitled his paper, “The Capricious, Actively Evasive, Unsustainable Nature of Psi”. This title suggests the frustration that the neoparapsychologists experience. It seems as if the findings are teasing the researchers. Indeed, Kennedy’s preferred hypothesis is that some intelligent agent is somehow deliberately manipulating the evidence so that psi appears to be real but always evades being captured by the scientific method.
The neoparapsychologists discern patterns in the parapsychological data. In many cases a new line of research contains a mixture of results that include some insignificant outcomes but also many significant results. With the passing of time the significant outcomes become fewer until finally the average effect size approaches zero. Another pattern is where the results, instead of going to zero, actually change from psi-hitting to psi-missing. Of course, there is the experimenter effect: some researchers consistently get positive results (but only for a time); some consistently get negative effect sizes; and others do not get significant outcomes. The problem is that these patterns are unpredictable and cannot be captured by scientific methodology.
The neoparapsychologists do not just admit that parapsychology has failed to produce evidence that can scientifically pass scientific muster. They go further and argue that this ability to evade scientific scrutiny is an inherent and unique property of psi. Psi exists, they maintain, but it will require a new kind of science to recognize it. Jahn and Dunne (2008) call for changing the rules of science to allow psi to be accepted. This obviously begs the question that they assume psi exists. But they cannot prove it by current scientific methods. Therefore let’s change the methods to allow psi to enter the halls of science.
Because they do not doubt the existence of psi, the neoparapsychologists seek explanations for why it manifests itself in such quirky ways. Most seem to prefer analogies with the seemingly odd behaviour of quantum phenomena. Kennedy, as indicated, hypothesizes that some unknown intelligence is deliberately teasing the researchers. Such a hypothesis, which takes us back to the mischief of the ancient gods, is still another indicator of how far parapsychology is straying from the world of science.
These attempted explanations of the apparent pattern of results are little more than re-descriptions of eccentricities in the obtained results. Although the parapsychologists try to find evidence for a coherent phenomenon they label psi, they have never managed to propose a positive definition that would enable them to predict and detect the presence of psi in the data. Instead, they rely on a negative definition to detect the presence of ESP and PK. They declare psi is present whenever the outcomes vary significantly from a chance baseline and no mundane explanation is readily at hand. An example of this was discussed in my discussion of effect sizes in meta-analyses.
Alcock (2003) provides a relatively complete list of the major problems faced by parapsychology. Here, I will focus just on the lack of a positive concept and test for detecting the presence of psi. One consequence is that the parapsychologists have proposed many sufficient conditions for the presence of psi but lack any necessary conditions. For example, Rhine and other early parapsychologists made much about the discovery of the “decline effect” within parapsychological experiments. This decline effect was hailed as proof of the existence of psi, even in experiments where the overall effect size was zero. However, when the decline effect was not discovered in other experiments, this did not prevent the researchers from declaring the presence of psi if they detected some other pattern that differed from chance. This creates the unsatisfactory situation where a wide variety of patterns can be used to demonstrate the presence of psi, but there is no way to demonstrate the absence of psi. This, by itself, can contribute to a large number of spurious successes. And, of course, it makes the claims for psi unfalsifiable.
Another defect of this negative approach to the detection of psi, is that we have no disciplined way to claim that an effect size in one experiment is due to the same cause as the effect size in another experiment. I have already discussed how the neoparapsychologists create abstruse explanations to account for the peculiar patterns in the research data. These explanations assume that the various patterns all result from the bizarre and impish behaviour of a single, coherent phenomenon they call psi. But it seems much more likely that the different patterns such as the experimenter effect, the decline effect, and the reversal of effect are due to a variety of different phenomena. The lack of a positive definition and test for psi does not allow the parapsychologists to rule out this possibility.
Moreover, the claim that these evasive patterns are unique to psi research is just not true. Throughout the modern history of science, individuals have claimed to have discovered anomalies which challenged the specific scientific programme within which they were working. Some of these claims, on further investigation, turned out to be based on solid, replicable evidence. These resulted in appropriate accommodations to the relevant theory. Other such claims ultimately could not be consistently replicated and now occupy the scrap heap of science. The proponents of these discredited claims defended the failure of consistency and replicability with arguments that resemble those of contemporary parapsychologists – experimenter effects, decline effects as properties of their claimed phenomena, reversal of effects, etc.
Conclusions
If the neoparapsychologists are correct, then parapsychology, as it was envisioned during its first 160 years, is dead. To the extent that parapsychologists continue their endeavours, they will do so without a phenomenon that can be scientifically demonstrated and measured. Hopefully, the parapsychologists and the scientific community can learn important lessons from this failure. The parapsychologists attempted to be scientific by using key components of scientific methodology, but scientific methods without a lawful, systematic and replicable phenomenon cannot be science. Because it is so obvious, philosophers and other commentators on science and scientific method rarely focus on the fact that most of the day-to-day effort of scientists is devoted to making sure that their findings are trustworthy, lawful, communicable, and independently replicable. This is the necessary ingredient. Without replicable phenomena, science cannot exist.
References
- Article by Ray Hyman: “The Demise of Parapsychology, 1850-2009“.
- Alcock, J.E. (2003). Give the null hypothesis a chance: reasons to remain doubtful about the existence of psi. In Alcock, J.E., Burns, J.E., & Freeman, A. (Eds), Psi wars: Getting to grips with the paranormal (pp. 29-50). Charlottesville, VA: Imprint Academic.
- Bem, D.J. & Honorton, C. (1994). Does psi exist? Replicable evidence for an anomalous process of information transfer. Psychological Bulletin, 115, 4-18.
- Bierman, D. J. (2001). On the nature of anomalous phenomena: Another reality between the world of subjective consciousness and the objective world of physics? In P. van Locke (Ed.), The physical nature of consciousness (pp. 269-292). New York: Benjamins.
- Broughton, R.S., & Alexander, C.H. (1997). Autoganzfeld II: an attempted replication of the PRL ganzfeld research. Journal of Parapsychology, 61, 209-226.
- Honorton, C. (1985). Meta-analysis of psi ganzfeld research: A response to Hyman. Journal of Parapsychology, 49, 51-91.
- Hyman, R. (1985). The ganzfeld experiment: a critical appraisal. Journal of Parapsychology, 49, 3-49.
- Hyman, R. (1994). Anomaly or artifact? Comments on Bem and Honorton. Psychological Bulletin, 115, 19-24.
- Jahn, R.G., & Dunne, B.J. (2008). Change the rules! Journal of Scientific Exploration, 22, 193-213.
- James, W. (1960). The final impressions of a psychical researcher. In G. Murphy & R.D. Ballou (Eds.), William James on psychical research (pp. 309-325). New York: Viking. (Original work published 1909)
- Kennedy, J.E. (2001). Why is psi so elusive? A review and proposed model. Journal of Parapsychology, 65, 219-246.
- Kennedy, J.E. (2003). The capricious, actively evasive, unsustainable nature of psi: a summary and hypotheses. Journal of Parapsychology, 67, 53-74
- Lucadou, W.V. (2001). Hans in luck: The currency of evidence in parapsychology. Journal of Parapsychology, 65, 3-16.
- Radin, D. (1997). The conscious universe: the scientific truth of psychic phenomena. San Francisco: Harper Edge.
- Sidgwick, H. (1882). Presidential Address. Proceedings of the Society for Psychical Research,1, 7-12.
- Utts, J. (1995). An assessment for the evidence for psychic functioning. Journal of Parapsychology, 59, 289-320. [Also available at http://anson.ucdavis.edu/~utts/air2.html]
