The Variety of Stimulus Errors in Phenomenology

Rubin’s remarks on the role of language are similar to Bozzi’s discussion of the variety of stimulus errors (1972). Like Rubin, Bozzi claims that the stimulus er­ror stems from the fact that the same language is used to denote mechanical and projective properties as well as perceptual properties of objects. In experi­mental practice, this induces the belief that the research aims at comparing these domains in controlled conditions in order to replace the phenomeno­logical description with a description of what is known or hypothesized to occur in at least one of the links of its underlying causal chain. However, sup­pose that “square” denotes a perceptual form and the geometric property of the distal or the proximal stimulus with which the former should be compared and replaced. In reality, the geometric property is the result of measurement operations carried out on a perceivable object by means of instruments like a ruler or a pair of compasses. It is not a property of the stimulus that the percep­tion represents in the mind of the subjects. The stimuli are not the referents of perception, but rather the constructs built from mechanical and geometric measurements that enable one to design suitable controlled conditions to let the phenomena of interest occur and be put under scrutiny. Bozzi emphasizes that at a certain observational scale, the mechanical or geometric properties may be coincident with the perceptual properties. Yet they remain epistemo­logically distinct, for the latter may vary and the former remain constant only if the phenomenological factors and rules change.

Brunswik (1933) had remarked earlier on the epistemological role of this observable coincidence for a restricted set of features such as shape, size and speed. Bozzi claims that once the mechanical and geometric properties are unequivocally specified as operationalist properties, this coincidence can be exploited to define an abstract system of coordination between a selected set of perceptual properties and a set of mechanical and geometric properties. Thus these latter can be introduced in the experimental design. If the logical nature of this coordination is neglected, the stimulus error is committed. Bozzi (1972) builds a table of the variety of potential stimulus errors that arise if the construction of the experimental conditions and the measurement operations are assumed as posits that reify the external causes of perception (see table 1).

In the columns, DS stands for the distal stimulus; sc stands for the stimu­lus constellation, namely the energies and the form of transmission, which is studied by optics, acoustics, chemistry, conveying the stimulation from the dis­tal stimulus to the peripheral receptors; PS stands for the proximal stimulus; and $ stands for appearances. The processes, which are supposed to underlie perception, are represented by the concatenation of symbols DS, sc, PS that designate any relevant event in the space and time ruled by the laws of phys­ics and their restriction to physiology. In this sense, ascribing to one of these processes the role of explanatory units of particular phenomena means com­mitting some sort of stimulus error. The symbol “+” means that a perceptual feature is ascribed to that physical or physiological property, namely that a stage of the process that is known or solely conjectured to underlie perception is considered an explanatory constituent of perception, regardless of its being observable or not. This amounts to reducing or explaining away phenomena by the constructs and findings of other sciences. The table represents the pos­sible stimulus errors of the theory of perception, of which only some instances will be exemplified.

The first row represents a trivial case in which, for instance, it is argued or implicitly assumed that a thing appears to be square because the subjects know that it is physically a square and this property is somehow preserved through the stages of the causal process of perception. It is worth noticing that this error amounts more to neglecting that the theory needs to recognize the questions about perception as autonomous issues and to carry out a proper decomposition of perception to formulate and solve its own problems, than to a scientific explanation. This is the case for common-sense assumptions, in which naive realism is associated with the causal account of perception to which popularized scientific notions are applied, which has been critically considered by Kanizsa. The second row represents the explanation of a per­ceptual feature through knowledge about the proximal stimulus, because the same kind of feature could be produced in artificial conditions that physi­cally do not have it. For example, Ames’s trapezium in a particular orienta­tion is seen as rectangular, but the stimulus error prevents recognition that this perception depends on phenomenological factors (cf. Canestrari, 1975). The fourth row represents most accounts of the optical-geometric illusions. If measured, the lines of the Muller-Lyer figure are known to be of equal length and this property must be projected from the distal to the proximal stimulus; hence, the appearance is explained as the result of an erroneous interpretation of the physical distal or sensory property. The fifth row represents the accounts of perceptual constancies, which assume that the property of the distal stimu­lus is lost in the proximal stimulus so that it has to be recovered by an inference or past experience on the grounds of some knowledge base for it to appear.

Source: Calì Carmelo (2017), Phenomenology of Perception: Theories and Experimental Evidence, Brill.