Bozzi (1989, 2002) presents the theoretical and empirical arguments in support of experimental phenomenology as the science of perception whose domain is closed in the sense that concepts, constants and variables are admitted only if their meaning and course value is bounded by observable referents. However, the primitives of the theory have to be connected and its constructs reduced to the description of phenomena by means of a consistent and well-defined series of operations.
Phenomenological description plays a substantial theoretical and empirical role. The description of a perceptual scene may be either phenomenological or causal. In the first case, the description must capture the phenomenal units into which the scene can be meaningfully decomposed. To be sure, many descriptions can be given of the distinguishing units for a chosen level of specification. Likewise many descriptions may sort differently the features of the scene. Yet, if the truth of the descriptions is to be preserved, the descriptions have to be formulated and connected according to the way they express the directly perceivable content in the scene. In fact, there are many already available terms to describe the scene. Otherwise, if, for instance, one needs a technical vocabulary, the descriptive terms can be built from scratch. At any rate, even if description consists of connecting terms selected from lists of possible items, their choice and articulation will be dependent on the extent to which they denote the relevant observable feature and units of the perceptual scene.
Although he does not quote it, Bozzi seems to make reference to the notion of “word-world direction of fit” (Austin, 1953), according to which calling a triangle “square” is an error that violates language, but wrongly describing a triangle as a square fails to satisfy the facts (Anscombe, 1957; Searle, 1975). Indeed, Bozzi (1961a, 1968) had argued earlier that the condition of satisfaction of phenomenological descriptions depends on fitting the perceptual facts. There can be divergent opinions on the appropriate use of descriptive terms, but the question can be decided according to the extent to which a term makes the structure of the perceptual scene manifest to every subject. Even a deliberately false description is a borderline case of the word-world direction of fit, since one who wishes to deceive other subjects must actually be aware of what is perceptually the case in order to give a mistaken description of it.
The descriptions that satisfy the word-world direction of fit need to be composed of terms denoting observable units, be they things, parts or qualities thereof. Bozzi (1989: 15) claims that “observable data” refer to two kinds of entities: (1) perceivable objects, whose appearances undergo changes taking place under observation, because controlled transformations are induced in them and their results are directly observed; (2) the results of interconnected operations that are constructed as measures by applying particular instruments on defined regions or points of perceivable objects. The domain of phenomenological descriptions consists of observable entities of the first kind, which have particular epistemological properties (Bozzi, 1961a, 1976, 2002). Perceivable objects are not subjective in the sense of private data that reproduce material things inside each subject’s mind. Sensory or mental data like after-images, scotomas, mental images and thoughts show an immediate, complete dependence on the subjects. Instead, appearances of the outer world display a high robustness and steadiness across arbitrary variations depending on perceivers’ cognitive and motor behaviour. Perceivable objects are stable rather than unpredictably varying across circumstances and subjects. They hold relations to one another in the larger context of perceptual scenes, but also show various observable forms of independence of one another that cannot be altered by subjective decisions and behaviour. Indeed, ordinary experience attests that there are wide regions of the perceived world that are successfully treated as public, that is to say accessible to and ascertainable by all perceivers, rather than as private views in one’s own subjective experience. In reality, the perceptual world is so stable that it supports the interpersonal agency of subjects.
Bozzi claims that these epistemological properties are neglected if perception is considered merely as a means to provide a sort of psychological link between the subjects and the physical objects that exist “beyond” perception. In this connection, perceptual appearances are construed as signs or representations of objects that by definition cannot be attained, thus introducing the view that appearances are inaccurate if not deceiving about the real physical objects. Instead, if this representational interpretation of the reference of perception is rejected, a firm observational basis is derivable from appearances. On the one hand, perceptual appearances are the elementary constituents of the direct experience of the world, if “to appear” is interpreted simply as “to be present to,” “to be perceivable” rather than “to represent” a reality beyond perception. Perceptual appearances are not the seeming of non-observable correlates, rather “they are what they are.” Bozzi (1976, 1989) quotes this expression from Tolman (1951: 96) to show that even one who is not committed to phenomenology can recognize that immediate experience contains “enough objectivity” upon which both the physical and the mental non-perceptual constructs are built. On the other hand, if defined operations are carried out directly on perceivable objects, the observation of their effects hic et nunc allows deriving the units of analysis. For example, let two pieces of cardboard be cut from the same red sheet of paper. If the borders of these cardboards are cut so that one has a straight and the other has a zigzagged contour, then the red spreading on either of them appears with different lightness, saturation and mode of appearance (Kanizsa, 1954). In order to experience the appearance of the same red, it is necessary to look at the two cardboards through the hole of a reduction screen. The observable transformation of the borders makes the margins and the color appearances emerge as the features whose correlation requires investigation. Thus the transformation specifies the features and sets the condition of satisfaction of their phenomenological description. Let the Muller-Lyer figure be presented and the angular sectors be continuously modified, so that the apparent length of the in-between lines varies accordingly. The segments are seen to coincide in length only if the angular sectors come to hold the same position in relation to them. The perceived co-variation between perceived angles and perceived lengths must figure in a phenomenological description, thus allowing one to extract the repeatable units that can be assumed as the experimental variables that account for this observable fact.
On such epistemological and methodological grounds, Bozzi rejects the argument that phenomenal data are unobservable because no one can observe what falls in another’s perception. In addition, he proposes an operationalist refutation of this claim (1961a; see 1976 for the importance of Bridgman’s philosophy of science for perception). Since there is no way to test the meaning of a sentence like “no one can observe that no one can observe what falls in another’s perception,” the question whether phenomenal data are observable is meaningless and the argument is invalid. The question is similar to asking whether or not the change in the absolute scale of magnitudes as the solar system moves could be detected, since it is by hypothesis extended to all physical objects to the same degree (Bridgman, 1927: 28). Bridgman says that if the length is defined by the application of a measuring rod, there is no operation to answer the question. The question could have a meaning only for someone outside and above the solar system who defined the length with different operations from ours, but then the meaning of the concept of length would also be different. Indeed, Bridgman (1927: 30) says that asking whether the sensation that one calls “blue” is really the same as that which another one calls “blue” is an example of a meaningless question. Likewise the sentence quoted by Bozzi could have meaning only for the imaginary point of view of a subject over and above the perceiving subjects. Instead, the phenomenal data are the observables that emerge as ascertainable effects for given transformations of perceivable objects and provide the regions and the points upon which measurement gauges are applied.
Suppose we compare the meaning of sentences about the Muller-Lyer figure (1976). The sentence “I see two lines of different length” is true, for the sentence “I see two lines of equal length” is false. This sentence is different from “the two lines have equal length,” whose truth depends on transposing a pair of compasses from one line to the other and observing that the angle of the arms does not need to be changed for the points to coincide with the line endings. This operation shows that this sentence is true of the metric property of length. Instead, the former sentences have a meaning that derives from the operation of changing the size of the angular sectors, while their opposite truth- values derive from the perceptual effects it brings under scrutiny. The same holds for the controlled transformation of cutting the borders of two squares from the same cardboard with different margins. This operational definition is useful because it allows one to expect that the observable data of the perceptual and metric length will be coincident for those values of the inclination and size of angular sectors at which the transformations and the measurement operations converge. Suppose that subjects are asked to pretend to doubt their perception or knowledge of the Muller-Lyer figure according to the sentences: “I doubt whether these lines appear to have different length” and “I doubt whether these lines have equal length.” Bozzi (2002) claims that pretending to doubt whether the lines have equal length makes sense, if the corresponding sentence is uttered at the same time that a pair of compasses is transposed to them. It is possible to imagine that the length of the arms is not preserved across the translation because it is possible to test by repeated measures that it is not the case, and accordingly the truth-value of the sentence on the metric length is decidable. Instead, pretending to doubt whether the lines appear to have different length is meaningless, if the corresponding statement is uttered at the same time that the figure is under observation. There is no direct operation by which it is possible to discover that the perceived length could be otherwise at the same time the figure is seen. Indeed, pretending to doubt its appearance cannot induce one to imagine any change in the perception of the figure. Therefore, the evidence of this kind of observable data is undisputable so that the repeatable units of analysis can be abstracted from it.
Bozzi (1976) calls “Cartesian propositions” the sentences expressing the evidence of perception. They are descriptions of appearances that “are what they are,” because the meaning of the composing terms obeys the conditions of satisfaction set by observable transformations. They denote the objectivity of the perceivable objects to which the experimental protocols should likely approximate. This objectivity is not coincident with a postulated physical entity that appearances inadequately represent. It is only by means of the application of a double ruler that one is allowed to state “x = x” about the two lines of the Muller-Lyer figure under the respect of length. If repeated at will, this measurement operation allows us to construct the concept of the physical length but not to posit the length as an entity lying beyond the appearances.
The causal descriptions of perceptual scenes are composed by observable data obtained by means of measurement operations carried out through instruments whose theory is known. The knowledge obtained about the frequency of reflected light radiations, the optical and anatomical elements of the eye that exploit the wave-like and the particle-like behaviour of light and so forth up to the visual cortex neurons is used to construct the non-phenomenal correlates of the causal account of perception. Making causal descriptions does not put the epistemological value of the phenomenological description into question. Indeed, the observable results of measurement require that subjects agree on the regions and points of the perceivable objects to which the gauges have to be applied. Therefore, measurement operations are an indirect proof of the stable and robust character of phenomena.
The domain of experimental phenomenology includes both phenomenal data and Cartesian propositions and is characterized by the principle of “the co-planarity of variables” (Bozzi, 1989: 48, 1985: 28). Physicists are interested in the referents to which the observable data obtained through manometers, Geiger counters and thermometers point beyond themselves. Empirical sentences refer to the observable results of the operations, while the protocols refer to what these results are believed to be an indicator of. Instead, phenomenal data and propositions of experimental phenomenology are not indicators of something beyond them, because they capture the objective and repeatable content of appearances. From phenomenal data, perceivable experimental variables are extracted; hence they vary over the common plane of possible appearances (1993: 188). In such a case, empirical sentences refer to the perceivable values the variables take in the course of the manipulation in controlled
conditions, while the protocols are the intended interpretation of the Cartesian propositions. The chromatic induction exemplifies the sense of the variables’ co-planarity. Let a small grey square be laid on a larger blue square. The latter will induce a yellow appearance on the grey square. Next, let a half-white and half-black Maxwell disc rotate at the fusion speed in order to yield a grey appearance, and punch a hole in the blue square so that the rotating disc is visible through a hole. At suitable light conditions, the grey of the disc displays the same yellow as the grey square. The substitution of the grey square as one variable does not alter the yellow appearance, which is instead obtained as a repeatable datum, for the perceivable relation grey-blue is preserved regardless of the material objects.
Bozzi (1989: 26) claims that repeatable phenomena are the laboratory material of experimental phenomenology just as specimens of substances and crystals are for chemistry and petrography. Terms to denote the repeatable data are introduced in descriptions through ostensive definitions so that the experimental results are used to build a theory of perception that uses a minimal vocabulary of its fundamental elements and rules. Bozzi (1968: i69ff.) is aware of the objections against ostensive definitions. Using ostensive definitions means that the terms that have the semantic property of names are mentioned in a sentence like “This is N,” where “N” stands for the term to define, while at the same time a single perceptual object or quality is pointed at or exhibited. “N” can be either an individual or a general term, and the individual term is not restricted to being the subject of a sentence. Ambiguity and vagueness may arise, because it is not always clear which is the relevant level or dimension that is intended for N. For instance, in a perceptual ostensive definition, it might not be clear whether N requires pointing to a thing, one or more parts or qualities thereof, or its spatial and temporal extension. This problem may be solved by employing many successive ostensive definitions pointing at the same place but at different times, provided the spatial and temporal extension of the pointing or of the procedure followed to exhibit something are correctly distinguished from the extension of what is pointed at. Quine (1961: 67k) maintains that if one points to “a, b, c, …” as the successive referents as one defines “N,” one can understand the belongingness of “a, b, c, …” to the same x if at each ostensive definition the number of alternative candidates to “N” decreases so that x is the simplest object for “N” of which “a, b, c, . ” are parts. The meaning of “N” corresponds to what “a, b, c, .” have in common, rather than to their differences. Thus “N” is bound to refer to a single object rather than to manifold independent qualities.
Bozzi extends Quine’s solution to the production of phenomenological descriptions. The difficulties in fixing the reference of ostensive definitions are overcome because the perceivable referents emerge through repeated transformations, for example varying systematically Wertheimer’s grouping factors. If a row of aligned equidistant dots is given, the alignment can be broken while the distance between the dots is kept constant. The terms denoting curves can be ostensively defined, because of the evident difference between perceived curved and straight arrangements. If three dots close to one another are displaced so that angles appear, another term can be ostensively defined. Another ostensive definition is obtained if four dots are displaced to form a square. Starting from the same row of dots, a series of transformations can make alternative groupings and forms perceivable; hence, it is possible to introduce the inclusive disjunction “V” for the appearances that occur through the various displacements. If the even dots in the row move so that they approach the odd dots, while the mutual alignment is preserved, then “being a pair” can be defined. Once the terms for the distances and their variations have been defined, other terms denoting the decrease of the distance between the dot pairs can be introduced.
By means of the recursive application of these kinds of transformations and the ostensive definitions, the minimal vocabulary of the theory can be compiled, which could be further enriched by appropriate combinations of primitive terms. Ostensive definitions are anchored to the observable data. Every new term is introduced at each further transformation, so that the definition of a new term implies reference to observable data that had been defined earlier for simpler conditions. Thus the terms also denote order relations, like “less or greater than,” space and time, identity and causality.
Therefore, experimental phenomenology does not build a linguistic representation or a taxonomical classification of appearances. It aims at discovering the “equations of state” that account for the features and structures that explain possible perceptual scenes. The features correspond to primitive terms. The structures are articulated in the form of functional connections between terms according to the scheme x = f(y, w,…). If p and q are elementary Cartesian propositions, the transformations of the perceivable variables may allow discovering the relation R(p, q) that is also a Cartesian proposition. If other experimental conditions allow observing not only that q varies at varying p, but also that the variation of q depends on p, this dependence is expressed in the functional connection q = f(p). If a functional connection between the perceivable variables is obtained, then it may be concluded that it represents the rules of perception in the phenomenological language of the theory. The scheme of the functional connections has a descriptive as well as an explanatory role. If it figures in a well-formed string of the language of the theory, it states the lawful dependence between types of observable data. If it is present in the terms that are correctly combined in a phenomenological description, it is the true interpretation of a structure of the perceivable world. Consider for example the figure 18.
The subjects see the polygons that overlie the black rectangles, but also the numerals that are delimited by the outline of the polygons. The phenomenal transparency makes the stratification occur so that both the overlying transparent and the underlying black surfaces are perceived as a whole. However, if the contour belongs to the transparent polygons, it cannot appear at the same time as the boundary that segregates the numerals from the black rectangles. This appearance is explained if it is reduced to the rules that account for the inversion of the boundary in the so-called multi-stability. The perceptual state of affairs can be expressed by a series of functional connections in the form of material implications (modified and adapted from Bozzi, 1989: 36):
- continuity with the ground (or: underlying figure) D ground (or: underlying figure) appearance (Rubin, 1921);
- continuity with the figure D figure appearance (Rubin, 1921);
- figure D overlying appearance on the ground (Rubin, 1921);
- ground D underlying appearance beneath the figure (Rubin, 1921);
- closed region D figure appearance (Wertheimer, 1923);
- region delimited by parallel boundaries D figure appearance (Morinaga, 1941);
- region in contact with the figure boundary D ground appearance (Rubin, 1921).
The implications a, b, c, e, and g satisfy Metelli’s conditions for the perceived transparency of opaque co-planar two-dimensional surfaces and account for the first perceptual case. The implication d accounts for the second one. Of course, it has to be combined with other functional connections to explain the difference of the figure/ground relation with respect to the first case, but it is sufficient to make the direction of the common boundary reverse for the numerals to appear on the transparent layer that is completed behind them. Therefore, this perceptual case is an instance of the relation “p V q.” These two groups of implications are distinct strings of the theory that are at variance, which accounts for the alternation between the perceptual cases’ appearances. At the same time they describe the evidence of a perceptual datum. No one can see the stratified figures and the numerals simultaneously, although transparency is still seen if the numerals appear. To account for this observable fact, the theory needs to admit the inclusive disjunction in order to distinguish this kind of case from the proper inconsistency. For example, a perceptual case like the well-known Rubin invertible figure that could violate the asymmetry of boundary requires the exclusive disjunction “p + q,” which means “p or q, but not both.” In such cases, the two alternative appearances are mutually inconsistent, because no feature of the first appears in the second when the switch occurs between them. In contrast to figure 18, the perceptual alternation is now explained by the fact that two sets of implications are inconsistent and describe the alternating prevalence of competing appearances over one another.
The Cartesian propositions can be concatenated to reduce experimental data to known rules or to discover new rules. Bozzi (1985) reconstructs the logic of this procedure using Wertheimer’s grouping factors as example. Wertheimer starts from inducing transformations on a row of equidistant dots. If their distance is altered through translation, the nearer the dots are the more they are grouped together and segregated from the visually more distant dots. At decreasing distances, the perceived unification increases. Therefore, the formula <$u, u = fTJ can be introduced to express that the unitary form $u appears according to the proximity factor F1. One can hypothesize that this factor explains the grouping for each set of visual elements that satisfy proximity. In the figure 19, this factor prescribes that the angular sector on the right should appear as a unitary form against the line that could be formed by the remaining four dots on the right (see also Wertheimer, 1923: 320).
This expectation is not fulfilled, because two partially overlying angular sectors appear instead of a horizontal and two oblique lines. Then the formula Φu, u = ƒ( F 1 ) ⋅ ( F¯ 1 ) can be written. There must be another factor that accounts for this observable fact and it can be expressed as follows: Φu, u = ƒ( F 1 ) ⋅ ( F¯ 1 ) → ( F 2 ) , where F2 is the term for the continuity of direction. This reasoning is repeatable for any perceptual counterexample of Fi, so that the recursion Φu, u = f ( F 1 , F 2 , F i , F n−1 , F n ) ⋅ ( F¯ 1 , F¯ 2 , F¯ i , F¯ n−1 , F¯ n ) → ( F n+1 ) is admitted. At each stage of the recursive definition, the proof of the new factor has the same form. Therefore, one can compile a list of all the features that are directly observable in a counterexample with the proviso that no feature is listed at the re-stage that has already been classified as a factor in the n _ 1 stage of the definition. Each feature provides a hypothesis on the new factor. It can be systematically varied and if its transformation causes the unified forms that appeared for earlier features of the list to change or to disappear, then it plays the role of a new grouping factor.
In conclusion, experimental phenomenology aims at discovering the logic of perceivable facts or of possible appearances. On the grounds of the experimental evidence of functional connections derived from perceivable variables, it gives a logical representation of perceptual features and structures. If it is shown that a structure is also shared by appearances of ordinary experience, the perception of the outside world is explained for the rule that corresponds to that particular structure.
Source: Calì Carmelo (2017), Phenomenology of Perception: Theories and Experimental Evidence, Brill.
10 Aug 2021
11 Aug 2021
11 Aug 2021
11 Aug 2021
11 Aug 2021
11 Aug 2021