Utts’s framing of the AIR question, whether the data demonstrate “non-sensory access to information”, deliberately uses a description-neutral phrasing that focuses on the data pattern rather than on any specific mechanism. The phrase encompasses what the laboratory protocols were designed to test: under conditions in which sender, target, and receiver were physically isolated, with computer-controlled target selection, with double-blinded judging of receiver-produced descriptions against target pools, would receiver descriptions correspond to targets more often than chance allows? This is a description of an information-access pattern, not a commitment to any particular causal account (precognition, clairvoyance, electromagnetic coupling, statistical artifact, residual sensory leakage). Utts’s report consistently maintained this descriptive framing.¹

The SAIC experiments documented in the July 25 memorandum were, per Utts’s report, “all designed to answer questions about psychic functioning” rather than merely to add additional confirmation of existence. That is, the SAIC program had moved past the demonstration-research phase Utts described in her 1993 collaborative paper into hypothesis-testing research, with each experiment designed to address a specific question: target-pool design effects, sender presence/absence, distance effects, target-feedback effects, and so forth.¹ This shift toward hypothesis-testing was precisely what Utts had argued the field needed; she treated the SAIC corpus as the most methodologically mature segment of the laboratory remote-viewing literature available at the time of the review.

Utts’s central statistical conclusion in the AIR review was that “the effect sizes reported … are too large and consistent to be dismissed as statistical flukes.”¹ The argument rested on two observations: first, that the effect sizes were of magnitude comparable to small but accepted effects in medicine and psychology (where d ≈ 0.2–0.3 is treated as scientifically meaningful when replicated); and second, that the SRI and SAIC effect sizes converged closely (0.209 vs. 0.230) despite the two programs being run by partially different personnel under partially different protocols, a convergence that would be unlikely if either set were a chance artifact. She did not conclude that a specific psi mechanism was confirmed; she concluded that the data pattern was real and required explanation. The data/mechanism distinction was made explicit in the report.

The methodological argument Utts emphasized was that the SRI and SAIC programs, while operating under partially shared personnel and institutional culture, were nevertheless distinct enough to constitute meaningful cross-laboratory replication. The SAIC program had been designed in part to address methodological criticisms raised against the earlier SRI protocols, including criticisms about target-pool sensory-leakage and judging procedures, and the persistence of effect sizes at the same magnitude under the improved SAIC protocols was, in Utts’s view, evidence that the effect was robust to the specific artifact concerns raised against the earlier work.¹ Critics, including Hyman, accepted the convergence point but argued that SRI and SAIC were too closely related to constitute genuinely independent replication and that confirmation outside the SRI–SAIC institutional lineage remained outstanding.²

Following the AIR review, Utts published a formal response to Hyman’s report addressing each of his objections.³ On the mechanism objection, she returned to the aspirin analogy she had used in her 1991 Statistical Science rejoinder: clinical efficacy was accepted before mechanism was understood, and demanding mechanistic explanation as a prerequisite for accepting a data pattern is a standard not uniformly applied across sciences. On the independent-replication objection, she pointed to the ganzfeld and PRL meta-analyses as constituting independent laboratory programs producing comparable effect sizes, with the autoganzfeld results in particular addressing the specific artifact concerns raised against earlier ganzfeld work. On the data-handling audit objection, she acknowledged the limitation but noted that the AIR commission had not provided for such an audit and that the absence of audit-level scrutiny applied equally to many accepted small-effect literatures in mainstream science.