The Stimulus Error. Unobservable Posits and the Variety of Data in Phenomenology

Kohler (1929a: 162) states that the psychology of perception has to avoid the “stimulus error,” namely it should not confuse “the knowledge about the physical conditions of sensory experience with this experience as such.” This phenomenological point has epistemological as well as methodological impli­cations for perception science. The stimulus error derives from assuming that perceptual phenomena are explained only if they are reduced to the sensa­tions whose physiological correlates correspond to the physical properties of the stimulation. For the reduction to be successful, sensations need to be set in a one-to-one correspondence with stimulation ([1913a] 1971: 14). Kohler calls “constancy hypothesis” this assumption of the strict correspondence between stimuli and sensations. If there is evidence that in particular instances this is not the case, auxiliary hypotheses are added according to which the correspon­dence is merely not noticed because psychic functions integrate the sensations in order for the perception to occur.

Kohler does not deny that each science has a “theoretical structure,” namely a chosen set of constants that allow for identifying an initially unintelligible mass of data on which to impose an order by reducing them to its laws. He also acknowledges that it is reasonable that a science holds onto the established laws when it is not yet known whether observations will prove their validity, and may have “a sort of attic” in which to keep the data that do not yet fit the theory ([1913a] 1971: 13, 29). Nonetheless, the progress of science and in par­ticular of the psychology of perception depends on preserving the variety of observable data and on establishing a standard to decide whether the data that depart from the expected value of observations are either predictable or de­pend on an incomplete description of phenomena or falsify the theory. Yet for a theory that admits the assumption implying the stimulus error, the standard is equivocal and the observations are undecidable so that no evidence may count as a refutation of the assumption. Such a theory is committed to unob­servable posits like sensations or psychic functions as well as to the theories implied by the admitted physical and physiological constructs. Furthermore, the auxiliary hypotheses are formulated as additional interpretative criteria to accommodate the theory to the observations that deviate from the theoretical expectations, so that it is no longer decidable whether they falsify the theory or not. On the contrary, Kohler maintains that every science needs a complete description of the phenomena in order to preserve the variety of observations and keep from disregarding those that are at odds with the theoretical assump­tions, thus selecting only the data that best fit the theory.

Kohler concedes that the constancy hypothesis may even be true “under spe­cific conditions and within certain limits,” but he remarks that from such cases it is generalized even when there is evidence to the contrary ([1913a] 1971: 14). In some cases, it is replaced by a simple law of combination of sensations, like that for color mixtures, which predicts that the constant correspondence with stimuli is maintained for particular aggregates of sensations. Yet the con­stancy hypothesis may induce one to select constants that are inadequate to the phenomenal features, so that fundamental data are overlooked even when occasionally observed ([1913a] 1971: 28). Kohler cites his research into the fea­tures of tones (1910, 1911, 1913b). He does not renounce the investigation of the physical properties of sound and their manipulation through interference tubes, the Galton whistle or the Stern tone variator. However, this is a means to bring about the appearances to discover and specify their essential attributes by observing their contribution to the perception of sound without taking for granted preconceived opinions that stem from physics or musical theory. For example, he came to find that, as Brentano had already suggested, pitch and loudness are not the unique perceptual attributes of sound (1911: 98, 1913b: 1). There is another “abstract” part, namely another constitutive element or “at­tributive part” in the sense of Brentano and Stumpf, namely the vowel charac­ter of sounds that belongs even to simple tones (1910: 284-285, 1911: 79). On the one hand, high-pitched tones have the vowel character “i,” low-pitched tones have the vowel character “u,” while tones at intermediate pitches have the vow­el characters “e, a, o” in descending order. On the other hand, each vowel con­tains a specific tonal feature of a definite pitch that enables one to recognize the same vowel character in complex sounds of a whole different pitch. This phenomenal feature underlies the similarity between tones and vowels, but it also allows that a tone that does not sound like a particular vowel is perceived as an in-between tone depending on its similarity to the vowel characters of the neighboring tones (1911: 98). The phenomenal values of this feature are ar­ranged in orders that have distinguished points and that recur in the form of an octave scale (1911: 111f., 130, 1913b). Kohler emphasizes that this feature had already been occasionally observed and neglected because of the constancy hypothesis, although it can also be correlated with particular ranges of waves frequencies ([1913a] 1971: 28, 1911: 130, 1913b: ioof.). Indeed, findings of such re­search that does not commit the stimulus error can be used to test the validity of theories of sound perception that rest on Fourier’s theorem (1911: 63).

If a simple law cannot replace the constancy hypothesis, auxiliary hypoth­eses are usually introduced so that the evidence of a lack of one-to-one cor­respondence between perception and stimulation is interpreted as proof that either the underlying constant sensations occur even if unnoticed, or psychic functions alter their observation. An example of the first kind of auxiliary hy­pothesis is the account of hearing the single tones of a chord or the pitches of the overtones in a complex tone. If the physical properties of the stimuli are the independent constants to which sensations correspond, the difference be­tween perceiving a chord or a tone as well as a complex sound or its overtones cannot be a phenomenal difference. The sensations in both cases are by hy­pothesis the same. Nonetheless, if the attention is focused on the appearances of a chord or a complex sound, it is possible to single out the constant sensa­tions. The conclusion is drawn that the sensations corresponding to the stimuli are only unnoticed, that they are bound to remain unchanged and that the perceptual difference is in reality due to attention. As regards this kind of argu­ment, Kohler raises the objection that it cannot prove the hypothesis. Since sensations are by hypothesis unnoticed, they are not usually observed and can be found as already implied in any counterexample. Besides, the change due to attention should prove the constancy of sensations, but in fact it presup­poses what it should prove, because the attention reveals afterward that which by definition cannot but be observed as such, since it cannot appear at first blush. An example of the second kind of auxiliary hypothesis is the account of optical-geometric illusions like the Muller-Lyer figure. The phenomenal differ­ence of the two lines is ascribed to the error of judgement about their length, which is caused by the interference between the angular sectors and the visual extension of the lines. If this interference were removed, the judgement would be expected to grasp correctly the sensations of length that correspond to the metric length of the segments. The hypothesis that the difference is not per­ceptual but is rather due to a psychic function that makes unobservable the one-to-one correspondence between stimuli and sensations rests on an expe­rience that is invoked as an additional interpretative criterion. If the subjects are trained to compare objects in such particular conditions, the illusion due to the mistaken judgement may be reduced the more they succeed in isolating the lines from the angular sectors. The reduction of the illusion is interpreted as the emergence of the unnoticed sensations of length. This account implies that the judgement is also unnoticed, while only the error is experienced. If the lines are put one above the other and two vertical lines are drawn to intersect them at their ends, the illusion disappears and is replaced by the correct judge­ment of length ([1913a] 1971: 24, 31 n. 31). This construction, which allows the equal length to appear, is interpreted as the suitable condition for the unno­ticed judgement to emerge (for a phenomenological use of such a construction see infra §4.2).

Kohler ([1913a] 1971: 24) maintains that this auxiliary hypothesis cannot prove the constancy of the sensations and that the experiences presented as additional criteria are in reality ad hoc adjustments of the theory. Even granted the role of judgement, it may account for the unobserved constant value of sensations only if “it is taken for granted that sensations are strictly stimulus bound.” Likewise, the decrease of the alleged illusion due to training reveals sensations that are already presumed to remain constantly tied to the stimuli. Once again, the constancy hypothesis is already implied by the auxiliary hy­pothesis that should prove it. Even the construct of a judgement error is not scientifically sound if the conditions in which it should occur and the reason for weighing it against “the most careful” observation are not conceptually ana­lyzed. Indeed, the construct of the illusion due to an experienced error rests on the assumption that the judgement forces perception to undergo blindly an external influence. Thus, shouldn’t we “expect always to regard this judgment with a certain skepticism, even when it agrees completely with what is to be expected in terms of the stimulus”? ([1913a] 1971: 30). Moreover, if the constant correspondence is unnoticed by hypothesis, how is one entitled to trust the observation of the decrease in or disappearance of the illusion, instead of the observation of the lack of correspondence between the perceived and metric length? Finally, if the judgements that are supposed to alter the appearance of the correspondence are unnoticed, they may have a “great ability to adapt to all kinds of changing conditions” ([1913a] 1971: 28, 32). How could they then be dis­tinguished from other psychic functions that could influence perception? How to decide between competing explanations for each case if not arbitrarily?

Kohler concludes that if a theory admits unobservable posits, the questions become experimentally undecidable. The commitment to extraneous theories implied in the physical and physiological constructs and the constancy hy­pothesis are such to claim validity beyond the scope of observation. The auxil­iary hypotheses are a “bonus to maintain artificially” the basic assumption. The additional criteria do not minimize the influence of the auxiliary hypotheses on the interpretation of data, since they are not independent of what the latter are intended to prove. Accordingly, there is “no possibility of reaching a deci­sion between an explanation in these terms and trust in observation” ([1913a] 1971: 24). Since observations do not have a decidable meaning, the constancy hypothesis implies that the deviations from basic assumptions are neglected ([1913a] 1971: 27). Instead, if observations depart robustly from assumptions, this means that there are more phenomena than those admitted by the constants of the theory. However fundamental the basic assumptions may seem given the present state of knowledge, they are never sufficient to prefer the seemingly simple order they confer on data to the unpredicted variety of phenomena. Rather, this demands a more comprehensive description of phenomena and an unprejudiced exploration of their features. For example, Kohler remarks that the lines of the Muller-Lyer figure are not spontaneously perceived as iso­lated. Therefore, it is reasonable to think that its appearance derives from the properties of space perception. Thus, rather than implying the postulation of unnoticed sensations corresponding to the metric properties of the stimuli, it requires the study of the structure of visual space ([1913a] 1971: 29, 30). In con­clusion, Kohler’s criticism leads to the phenomenological claim that percep­tion science has to deal with the complex cases of “the everyday perception of things” rather than solely with the cases admitted by the scope of validity of the constructs of other sciences or by unobservable posits ([1913a] 1971: 39).

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