Meijer et al. (2012) state, “Interestingly, the University of Illinois patented the original P300 based CIT as published in Farwell and Donchin (1991). And conveniently, the ‘discovery’ and patenting of the MERMER liberates him (Farwell) from the constraints of this earlier patent.” This statement, besides being irrelevant to the scientific issues at hand, is unequivocally and demonstrably false. According to patent law, a prior patent takes precedence over any future patent. Everything in the prior patent remained unaffected by Farwell’s four subsequent US patents. To obtain additional patents, Farwell had to prove to the satisfaction of the United States Patent and Trademark Office (USPTO) that his new discovery of the P300-MERMER was “novel, useful, and non-obvious” over the state of the prior art, including the University of Illinois patent and the P300. What “liberates” Farwell from the previous patent was not his new patents, but rather the fact that the University of Illinois failed to pay the maintenance fee, so the USPTO ruled the patent abandoned.
Meijer et al. (2012) appear to advocate an absolute taboo against mentioning in a scholarly work anything that has not been previously published in a peer-reviewed journal. We disagree. Other types of publications are published because they are perceived to have merit and can provide relevant and useful data and insights. Patents require proof that the patented invention is “novel, useful, and non-obvious.” Book chapters and encyclopedia entries are scrutinized by knowledgeable editors. Doctoral dissertations pass muster with committees of experts. Conference abstracts have some measure of scrutiny by editors. In the context of Farwell (2012), Farwell’s relevant previous publications include not only six previous peer-reviewed scientific papers (Farwell 2011a; Farwell and Donchin 1988, 1991; Farwell et al. 1993; Farwell and Smith 2001; Rapp et al. 1993),Footnote 3 but also a dissertation (Farwell 1992), two book chapters (Donchin et al. 1986; Miller et al. 1987); an encyclopedia entry (Farwell 2013), five patents (Farwell 1994, 1995a, b, 2007, 2010), a legal publication (Farwell and Makeig 2005), a monograph (Farwell 2011b), and several conference abstracts (e.g., Farwell, Richardson, and Richardson 2011). Moreover, since Farwell (2012), Farwell and colleagues have published four additional relevant studies (see Farwell, Richardson, and Richardson 2012).
Among the authors of the two papers under discussion here (Farwell 2012; Meijer et al. 2012), Farwell is not the only one to include discussion of such sources in his scholarly writings. Meijer et al. (2012) cite and meaningfully discuss Farwell and Donchin (1986), a conference abstract. In book chapters authored by both Farwell and Donchin, along with others (Donchin et al. 1986; Miller et al. 1987) the authors discuss in considerable detail the methods, results, and relevant data on P300, memory, and aging published previously in Farwell et al. (1985), a conference abstract—and not published in full form in a peer-reviewed journal. Applying an absolute taboo against discussing such data only to Farwell (2012) and not to the writings of others—such as the authors of Meijer et al.—would be discriminatory and inconsistent, and would not serve the best interests of readers who would like to know the full story.
Moreover, fortunately, the discussion of the distinction between conference abstracts and peer-reviewed publications has become largely moot with the recent peer-reviewed publication (Farwell et al. 2012) of four studies cited in Farwell (2012) as conference abstracts, and will become entirely moot with additional upcoming publications.
Meijer et al.’s (2012) table 1 represents the peer-reviewed publications and “verdicts”Footnote 4 therein at the time of Farwell (2012). It does not, however, represent a complete picture of all of the relevant evidence on brain fingerprinting at that time, as discussed above. Moreover, since that time an additional four studies including 76 subject tests on 74 individuals have been published in a peer-reviewed journal (Farwell et al. 2012). (These were previously published as conference abstracts.)
Meijer et al. (2012) accuse Farwell of misrepresenting conference abstracts as full-fledged peer-reviewed publications. Consider the following. Farwell (2012) cited Farwell and Donchin (1986) as follows: “In the initial brain fingerprinting research, Farwell and Donchin used the P300 event-related brain potential (Farwell and Donchin 1986….” “Three types of stimuli are presented: probes, targets, and irrelevants. (Farwell & Donchin 1986…” “Farwell and Donchin (1986, 1991) made it clear that brain fingerprinting detects information, not lies, guilt, or actions.”
Meijer et al. cited the same publication as follows: “The variant of the CIT with ERPs was first investigated in the late 80ties [sic] (Farwell and Donchin 1986…” In the reference sections of the respective papers, the citations are identical word for word. Other citations in both papers are similar.
Meijer et al. (2012) provide no criterion by which they judge Farwell’s citations to be “misrepresenting” and their own to be clearly delineating the difference between conference abstracts and peer-reviewed papers. Our perspective is that the readers of Cognitive Neurodynamics are highly intelligent and knowledgeable. We presume them to be intelligent enough to follow Farwell’s discussion of the significance of intracranial recordings in the inferior parietal lobe/supramarginal gyrus, superior temporal sulcus, the amygdala and hippocampus, dorsolateral and orbital frontal cortices, and the anterior cingulate, and his discussion of the mathematical distinctions between bootstrapping classification and comparison algorithms and the resultant differences in statistical confidences. Such individuals, in our view, can find their way to the reference section and readily distinguish between different types of publications.
One clarification in terminology is in order. Meijer et al. (2012) point out a linguistic anomaly that might cause some confusion, and we would like to take this opportunity to clarify the situation. Consider the situation wherein John Smith is tested as a subject on a set of stimuli for which he is “information present,” and John Smith is also tested on a different set of stimuli for which he is “information absent.” How many information-present subjects are tested? One. How many information-absent subjects are tested? One. How many total subjects are tested? The answer “one” leads to the anomaly 1 + 1 = 1. The answer “two” is correct in terms of the number of tests run (and the statistical power of the design), but it is not quite correct in that two of the “subjects” were actually the same person. “Subject tests” may be a better term for avoiding ambiguity. The statement “There were two subject tests,” along with a disclosure of the experimental design in which the same participant was run as a subject in two different tests, provides a more complete and unambiguous account. The summary charts in Farwell (2012) (tables 2 and 3) used the word “subjects” to refer to the number of “subject tests” undertaken, even when one individual participated in more than one test. Substituting the column heading “subject tests” would be a useful change that would clear up any possible ambiguity.
Meijer et al. (2012) use the term “participants” to refer to individual human beings and “verdicts” to refer to subject tests. In our view “verdicts” is inappropriate, because brain fingerprinting does not deliver a legal verdict, but only detects information.Footnote 5 We prefer the term “subject tests” for reasons described above.Footnote 6 The term “participants” may also be ambiguous, as it may be construed to refer only to people who participated in a crime or mock crime, or to all participants in the research.
None of this is an issue for anyone familiar with the relevant literature, however, because in all of Farwell’s publications (e.g., Farwell and Donchin 1991), the number of tests and the number of people who participated have been clearly delineated, and the authors have clearly disclosed when one person is a subject in more than one test. Moreover, this makes no difference in the statistics computed or the scientific conclusions drawn from the data. Nevertheless, we are happy to provide a clarification as above.
Meijer et al. (2012) impugned Farwell’s motives and character, as follows. As is common in the field, Farwell and Donchin published their research first as a conference abstract (Farwell and Donchin 1986) and later as a full peer-reviewed paper (Farwell and Donchin 1991). Both Farwell (2012) and Meijer et al. cite and meaningfully discuss both of the Farwell and Donchin papers. Obviously, Farwell’s comprehensive tutorial review includes more detail than Meijer et al.’s brief communication in the discussion of these and other papers. Farwell includes tables 2 and 3, which present the number of subject tests in the various studies discussed (see above discussion on terminology). Meijer et al.—and not Farwell—added together the numbers of subjects in the various studies such that the numbers were duplicated. That is, when two publications [an abstract and a subsequent full paper such as Farwell and Donchin (1986, 1991)] reported on the same research, Meijer et al. counted the tests twice, thus inflating the totals for field and laboratory tests. Thus the totals Meijer et al. computed for laboratory and field studies do not reflect the actual total numbers of individuals or subject tests in the studies. Then, on the basis of their own misrepresentative addition—which does not appear in Farwell (2012)—and/or on the basis of some difference they postulate but do not describe in the manner of citing the respective papers (see above discussion), Meijer et al. accuse Farwell of “deliberately duplicating participants and studies.”
We will leave it to the readers of Cognitive Neurodynamics to form their own judgments as to whether either, both, or neither of Farwell’s (2012) and Meijer et al.’s (2012) writings constitute “duplicating participants and studies.” We choose not to speculate on the motives of our fellow scientists, so we will not address the question of whether Meijer et al.’s actions in this regard were “deliberate” or not. In any case, in our view, Meijer et al.’s impugning of Farwell’s motives and character does not advance the progress of science.
In any case, none of this changes the fundamental scientific issues at hand or the scientific conclusions warranted by the data. The progress of science is driven by research and data, not by words. Any way you name, rename, misname, parse, count, recount, miscount, discount, or don’t count the publications, people, and tests, the fact remains that all available data (including Meijer et al.’s table 1) are compatible with Hypotheses 1–3 and the 20 scientific standards proposed in Farwell (2012).
Meijer et al. (2012) made a number of other ad hominem comments about Farwell, his motives, character, subjective state, intentions, behavior, writing style, etc. In our view, further discussion of such matters will not serve the progress of science or the interests of our readers.
Meijer et al. (2012) state that Farwell and colleagues implemented standard 4 for “some unexplained reason.” Standard 4 specifies using situation-relevant (or crime-relevant) targets, rather than inherently irrelevant targets made relevant only by instructions. In fact, Farwell (2012) explained in detail their reasoning and the considerable value of situation-relevant targets in reference to the FBI agent study, devoting 536 words and one figure to this. Farwell et al. (2012) explained this in even more detail.
Farwell (2012, pp. 118–122) devoted 4,005 words to a comprehensive discussion of the functional significance, antecedent conditions, history, neurodynamics, physiological mechanism, and signal characteristics P300 and P300-MERMER. Meijer et al. (2012) quoted one sentence of this out of context, represented it as Farwell’s view of the P300, and criticized it as being inadequate. We agree that this sentence, or virtually any other single sentence from Farwell or any other publication, is an inadequate description of the P300.Footnote 7 We encourage readers to read Farwell’s full article.
Meijer et al. (2012) state that the “P300-MERMER… is unlikely to solve the problem caused by the lack of a one-to-one relationship between P300 and memory.” Neither Farwell nor, to our knowledge, anyone other than Meijer et al. has suggested that there is (or should be) a one-to-one relationship between P300 and memory, or considered the lack of such a relationship to constitute a “problem” to “solve.” Again, we encourage readers to read Farwell’s (2012) entire article for a comprehensive discussion of the P300 and P300-MERMER and their role in the detection of concealed information.