On the structure and function of memory

1. The memorization and actualization of functional relationships
follow the same schema

As was shown: it is precisely those relationships between P-percepts, known as "gestalt stimuli" that actualize the gestalt factors Dm and Dl, and thus provoke the perception of themselves, i.e. the perception of large and small brightness differences and large and small location differences. It is precisely these same relationships that also form the gestalt factors Dm and Dl, respectively.

In Part 1, a general schema for the actualization of a gestalt factor by its gestalt stimulus, valid for every hierarchy level, was presented (Fig.1-13). Here it will be reproduced with other symbols (Fig.4-3A). If there  is  the  relationship  K  between  the  functional  (and  possibly phenomenalized)  entities J and J', K operates as  the gestalt stimulus that actualizes the gestalt factor K. In this way, gestalt stimulus K is additionally experienced, and the entire system "J and J' in relationship K" is subjectively perceived.

Figure 4-3. Schema for the formation and actualization of a memory content (gestalt factor)

The forming of a gestalt factor follows the same schema, as shown in Fig. 4-3B: the relationships K between the functional entities J and J' are memorized, so that the memory content K is formed, which in turn operates as the gestalt factor K, as shown in Fig. 4-3A.

2. Phenomenological and functionological concepts of the memorized objects

Distinctions have already been made between explicit and implicit memory, or learning process. Here, these distinctions will be dealt with in more detail. It was earlier, just like today, a popular, if not the predominant, belief that we can memorize only what we experience by way of perception, feelings, thoughts and imagination. It is believed that part of what we have experienced must somehow be imprinted on our brains, it must lead to an "engram", it must leave its "mark". Accordingly, the brain should then at some later point in time - it does not matter how - relate what is experienced now to that which was previously experienced. This can either happen in the form of a "free memory" ("cued recall") (when one remembers the previous experience) or by way of a "bound memory" (when previous events influence actual experience without necessarily being remembered). I relate this kind of memorization to the "phenomeno- logical concept of the memorized objects". Its vital assumption is: what has not been experienced, cannot be memorized, and therefore cannot be remembered, and thus not influence further experiences. Fig. 4-4A shows this simple process: an experience P1 is the input of the memory, and becomes here the function F1. F1 will, when actualized, produce phenomenon P1, from where it was formed. This memorization of phenomenal entities is called an "explicit" learning process.


Figure 4-4. Phenomenological (A+B) and functionological (C) concepts of the memorized objects

The same process of memorization, with a more complex phenomenon to be memorized, is shown in Fig. 4-4B: two phenomena P1a and P1b are perceived, and the perceiver sees additionally the relationship P2 between these two phenomena, this situation is depicted below in the phenomenal sphere. If this complex phenomenon is memorized, the memory content "relationship F2 between F1a and F1b"  is formed, in the functional sphere. When this memory content is actualized, the previously established complex phenomenon occurs again, depicted above in the phenomenal sphere.

According to the ETVG, however, the sort of memorization that operates at the psychic evolutionary level (in the visual system, for example, at the Sublevels 1 to 10), is related to another concept, the "functionological concept of the memorized objects": not the phenomenal entities are memorized, but the functional entities; this is called an "implicit" learning process.  Fig. 4-4C shows this sort of memorization. The factor F1 is actualized twice: as F1a and F1b, and produces two qualities P1a and P1b (in the phenomenal sphere). Also the relationship "2" between the two states F1a and F1b (in the functional sphere) is memorized (in the functional sphere as well).  This process of memorization produces the new, i.e. previously non-existent, memory content "relationship F2 between F1a and F1b" (in the functional sphere), with the emergence of the new function F2. When actualized, this memory content behaves as a gestalt factor, and produces the percept "relationship P2 between P1a and P1b", with the new gestalt quality P2. The difference between these sorts of memorization is clear: in Cases A and B, an entity of the phenomenal sphere is memorized, in Case C an entity of the functional sphere. The evolutionary advantage of Case C over B (and A) is, that an absolutely new experience can occur (the experience P2), which has never occurred before, as it was non-existent in the phenomenal starting situation, where only P1a and P1b, but not the relationship P2 between them, were experienced (as assumed in Case B). In the Case C, a new phenomenal quality is born; in the Cases A and B, a given phenomenal quality is purely stored, to be called back into consciousness at some time in the future.

The difference between these two kinds of learning processes is not widely accepted in the world of psychology. Many scientists assume explicit memory to be the only one. When a baby sees a bottle of a certain form, for example, and cries: "Oh, there is a bottle, the same form as I remember having seen yesterday", then the experience "bottle of this form" (in the phenomenal sphere) is a "free memory" (cued call) produced by the explicitly acquired memory content (in the functional sphere). In this case, the baby can distinguish between both the present and the past existence of the bottle. When the baby, however, drinks each day from a bottle of a certain form, and can remember having seen a bottle yesterday, but cannot remember ever having seen this form of the bottle before, although it perceives this form now, then "this form" is a "bound memory", produced by new form factors that have been developed in the meantime by a hidden ("implicit") learning process, so that the baby is able to perceive form qualities which it previously could not.

These two sorts of learning processes do not exclude one another; they are just at different levels; the lower level is the "implicit memory" with the "explicit memory" being founded on it: before the baby can (explicitly) remember having seen the form of the bottle before, it must have (implicitly) memorized the general construction principles, from which particular forms can be made. (These principles will be dealt with in Part 5.)

The ETVG, which has been developed by using the functionological concept of the memorized objects, is open for the possibility of a phenomenological concept. In Fig. 4-4C the relationships between the functions F1a and F1b were assumed to have been memorized. This is true for the case that these functions were actualized and thus produced the phenomena P1a and P1b, as well as for the case that this actualization did not happen. If the actualizations are established, and one later finds the phenomenon P2, one is inclined to believe that the abstract relationship 2 between P1a and P1b, too, would be experienced, and thus also memorized. This opinion corresponds to the phenomenological concept of the memorized objects, whereas the ETVG still supports the assumption that it is the functional conditions of the phenomena's relationships, i.e. the relationships between F1a and F1b, that have actually been memorized,  as depicted in Fig. 4-4C.

Representatives of the traditional (the phenomenological) concept are not very open to other theoretical possibilities of obtaining learning effects. This is true also for the Gestalt Theory. According to which memory is unable to introduce an absolutely new kind of happening into a non-gestalt process (see Part 10).They claim that a configurated memory effect is conditioned by previous perceptual experiences of approximately the same configuration, as Fig.4-4B shows. This phenomenological concept of the memorized objects attaches only small importance to the memory itself, which is purely reproductive, not capable of creating anything new, and only preserving something. The memory, in the sense of the functiono- logical concept of the memorized objects, is productive and creative, it is in the true sense of the German term a "gestalterisches Gedächtnis" (designing, artistic, creative memory). This sort of memory creates innovations of a higher order, it is a driving factor of evolution, by autonomously forming a hierarchical system. Each of the 25 ETVG gestalt factors is a "novum" in respect to the factors, on which it is grounded. Traditional perceptual science is not able to develop a fruitful theory of perception, because it knows only one or two spheres of being: the material and - if at all - the phenomenal sphere, but not the functional sphere of being. Therefore it looks for material "causes" for the phenomenal configurations.

In contrast to the "explicit" learning process, an "implicit" learning process is an early beginning ontogenetic, hierarchically formed, process of stable memorization, not of experienced (but "uncon- scious" and thus not recallable) abstract (functional) relationships between percepts, and is the foundation for both the development of explicit learning processes and holistic effects on behavior (see Kleine-Horst 1994b).

This implicit learning process enables the visual system to produce the "detectors" for itself, with the help of which it recognizes the "features" of its environment, simply by memorizing the subjectively non-perceived relationships between the already perceived "features". Can one imagine a more ingenious method of adapting the world recognizing apparatus to the structure of the world? The gestalt factors represent themselves as learned basic hypotheses about the structure of the world. Additionally, they provide the categories, with the help of which this world can be described. There is no other possibility of informing another human about a conscious visual experience (and thus the appearance of the "world") than with the help of the gestalt qualities which are born of the gestalt factors. It is even unimportant whether a verbal or a graphical description is used. In a drawing, the drawer must work with exactly the same gestalt factors as the perceiver of the drawing, in order to communicate about the visually perceived world. Furthermore, there is no other possibility to inform somebody verbally about the visual qualities one experiences than using of terms that designate these qualities, produced by the gestalt factors. It is unimportant, to which language the terms belong; what is relevant, is that both communicators  understand  this  language, whether  it  is a natural or a technical language, the ETVG-symbols, for instance. In each case, there is a limited number of "words" with different meanings, as there are only 25 relevant descriptive terms to designate the 25 existent different visual qualities (or classes of qualities).

Fig. 4-4 shows the major difference between the phenomen- ological and the functionological concept of the memorized objects: as the arrows indicate in A and B, the information to be memorized flows from the phenomenal to the functional sphere of being, where the information is stored. In Case C, however, all the information flows from the functional to the phenomenal sphere, as the entities to be memorized are of a functional nature. The functions, the relationships between which are to be memorized, can, of course, be actualized and thus also create phenomena. However, what is memorized are neither the phenomena themselves nor their functional conditions, but the functional relationships between their functional conditions, according to the ETVG hypotheses on the forming of gestalt factors.

3. The particular visual functions of single cells as memory contents

In this Part 4, the gestalt factors at the first five levels, Pml to Fl, have been assumed to be derivable as memory contents. In Part 3, the same gestalt factors had been assumed to be corresponding to functions of classes of single cells, located at the first five neuro- biologically definable levels of the visual information processing, which exhibits experienced qualities (from the retinal ganglion cells to the area V4). If these assumptions are true, then the single cell functions are memory contents. Since a certain gestalt factor (or its functions) is formed as a memorized relationship between other functions, a function of a single cell (or class of single cells) is formed by memorized relationships between other single cell functions.

As the neuronal excitations (and inhibitions) are relayed through neuronal connections from cell to cell, these connections are the site of the assumed implicit visual memory. It does not have to be assumed that the connections are formed entirely by a memorization process, it is sufficient to assume that the existing neuronal connections become "stronger" due to the frequency of using them. If, in this case, the neuronal "information flow" is facilitated according to the strength of the connections, the memorized cell functions will be actualized, and thus create their specific gestalt qualities.

If this model is respectively applied to the other gestalt factors, i.e. the form factors (Part 5 and 6) as well as the eleven figure factors with d- and t-aspects (Part 7), the ETVG is capable of accounting for visual experience. In the ETVG the visual experience, in the phenomenal sphere of being, is described by gestalt psychological terms; the Helmholtz' structuralistic point of view is used, but limited to the functional sphere, especially to the Z-factors; on the other hand, Helmholtz' association hypothesis is reanimated, although in an adjusted form, and also related to the functional sphere. The hierarchy assumption of the neurobiologists in regard to the perceptual process is also accepted. Finally, the actual-genetic facts of the Leipzig Gestaltists are taken into consideration (here thus far as a "virtual" actual genesis, but in Part 6 as a "real" actual genesis).

If the gestalt factors are implicit memory contents, and if the gestalt factors are identical to the functions of certain classes of single cells, then these single cell functions are predicted to be implicitly  acquired  in  early  infancy. The memory-based plasticity of neural networks is well-known. The ETVG prediction of the memory- based plasticity of functions in early infancy does not exclude plasticity in later developmental stages, not even in adults.


It is assumed that relationships between functional processes  are memorized more, the more often they occur in immediate spatial and temporal vicinity (contiguity) to one another (Empiristic Development Hypothesis). From both this hypothesis and the general structure of the retinal image that comes from the optical projections of the "objects in their surroundings", a hierarchy of memory contents is derived. These memory contents are implicitly memorized functional relationships, progressively acquired in early infancy, and operate as the hierarchically ordered gestalt factors Pml to Fl, and their positive and negative functions, when actualized by the same, or similar, functional relationships that were previously memorized.

If this hypothesis on the progressive formation of a hierarchy of visual factors is true, then the ontogenetic development of the baby's visual abilities must follow the bottom-up development of the factors. It was shown, that the predicted development is recognizable in experiments with babies, at least in respect to the successive occurrance of the factors Gml, Ll, and Fl.  Furthermore, according to the empiristic hypothesis, every visual factor's ability will increase with time, i.e. with the consolidation of the respective memory content (gestalt factor) by further stimulation of the retina, i.e. by further occurrence of the relationships that are to be memorized. Experiments reported in the literature show that, indeed, the function sensitivities increase with increasing age.

It is shown, that the forming and actualizing of the gestalt factors follow the same schema.  As the gestalt factors are implicit memory contents, the difference between implicit and explicit memorizations is described, and the phenomenological and functionological concepts of the memorized objects are explained.

Excerpt from L.Kleine-Horst: Empiristic theory of visual gestalt perception. Hierarchy and interactions of visual functions. (ETVG), Part 4, III

Back to table of contents