Outside the United States
In the Matter of Carrot Neurotechnology, Inc., File No. 1423132
Regarding the FTC definition of reliable scientific testing, related to perceptual learning, stating that "Such testing shall (1) be randomized, double-blind, and adequately controlled; and (2) be conducted by researchers qualified by training and experience to conduct such testing." As an expert in perceptual learning with ~25 years of experience, I believe that the FTC overestimates the placebo effect and the experimenter bias effect in perceptual learning. This assumes that all training procedures can produce the desired learning effect, but this is definitely not the case. Defining a placebo condition is also a matter of scientific inquiry to understand the mechanisms of learning. The literature shows that perceptual learning in vision can be very strong, and can generalize to untrained conditions when training follows specific protocols. However, robust/stable/long-lasting learning does not always take place, and even when there is learning, it does not always generalize (see my review of the field: Sagi, Vision Res 2011). There are several examples where learning is not efficient, for example, if the task is too easy or too difficult, or when using roving stimuli (Adini et al JOV 2004; Yu et al, JOV 2004). I recall the effort we made to find learning in some specific conditions but failed (Dorais & Sagi, Vision Res 1997; Polat & Sagi, PNAS 1994). Thus, perceptual learning (at least in vision) is a special form of learning, or of brain plasticity, that is not too sensitive to the above mentioned placebo and experimenter effects. Maybe the most important control to have is for the repeated testing effect, which is covered by Seitz and his coworkers.