I. The figure factors as memory contents

1. The Empiristic Development Hypothesis

The terms "gestalt stimulus", "gestalt factor", and "gestalt quality" are vital for the three-sphere theory of perception (not only the visual perception). The gestalt quality is a phenomenal entity, a content of the perceptual consciousness, and as thus is not possible to be grasped intellectually, it is "incomprehensive", it can only be experienced.

The gestalt qualities are entities of the phenomenal sphere of being, gestalt stimuli and gestalt factors are entities of the functional sphere. Functional entities are not directly experienceable, they are entities whose existence as conditions of the phenomena has to be assumed when one wishes to understand, to a certain extent, the phenomena. Gestalt stimuli should be taken as relationships and relation-relationships. Previously the carriers of these relationships had not been given a name; most often they have been termed "functional entities".

The question what gestalt factors are like, has remained open. In Part 3, however, it was already shown that a gestalt factor (or a gestalt function) behaves like a neuron's function, as the general properties of a neuron consist with the general properties of a gestalt factor. It was shown, furthermore, that the lowest six neurobiologically definable stages of static, two-dimensional visual information processing correlate to the first six stages of visual perception, defined by the ETVG.

In this Part, it will be shown that a gestalt factor also behaves like a memory content. The conditions under which gestalt factors are formed as memory contents are very simple; they were discovered, by the very first research into memory, to be conditions for memorizing facts: the more often one hears or reads a telephone number, for example, the better one can memorize it. Memory is a function and thus is located in the functional sphere. If we want to experience what we have memorized, we must first transform the functionally existent information to a phenomenally existent form, a "perceptual experience", in the phenomenal sphere.  

Consisting with the two different kinds of spheres, at the psychic evolutionary level, the functional and the phenomenal sphere, there are two different kinds of entities that can be memorized: functional and phenomenal entities. The latter are that we use when we intentionally memorize a telephone number. In this case, we can later recall this previously experienced number into our consciousness. This "explicit" learning process leads to a content of the "explicit memory". It is, however, also possible to memorize pure functional entities: relationships between the (functional) conditions of (phenomenal) experiences, for example. When actualized, the memory contents, formed by this process, can lead to an experience, too. In this case, however, the transformation of a memory content to an experience has not the form of "remembering", as it was not experiences, i.e. contents of consciousness, that were memorized - only previous experiences can be remembered, i.e. can be re-called into consciousness. Instead, the previous exclusively functionally existent entities were memorized, and lead, when actualized, to absolutely new experiences that have never occurred before. This learning process is called "implicit learning", and leads to contents of the "implicit memory". The ETVG assumes exclusively such implicit learning processes for the forming of the gestalt factors.

According to the general ETVG "Empiristic Development Hypothesis", stated for the gestalt factors, the following assumptions for the formation of gestalt factors will be made:

  1.  The relationship X between functional entities becomes via memorization the memory content X. After which, memory content X is actualized by the same relationship X between the functional entities, and presents itself as  the perceptual experience X.
  2. The more often the relationship X occurs, the more it is memorized.
  3.  It is the relationships between the functional entities, that find themselves in direct spatial and temporal proximity, that will be memorized - according to the well known "law of contiguity".
  4. The following identifications are valid:

                relationship X = gestalt stimulus X 
                memory content X  = gestalt factor  X
                perceptual experience X  = gestalt quality X

For the application of these particular hypotheses, the following starting conditions, given the first time a newborn opens its eyes, are assumed to exist:

This general structure of the retinal image (shown in the case of the "moon in the sky") will now be memorized by the infant because it is, indeed, frequently presented. The appropriate memory contents will be actualized by the following stimuli of the same, or similar, structure, whereby the perception of the environment is relatively "veridical".

2. Implicit and explicit memory concepts in the literature

The early gestaltists were nativists; they believed in innate "internal forces of organization" when they demonstrated their multi- faceted "Law of Prägnanz". The empiricists, however, stated that the perceptual preference of certain configurations would be a result of learning: those configurations would be preferred, to that, in the past experience, the subjects were exposed very often, compared to the less preferred configurations, which were seldom presented. The gestaltists tried to reject this assumption by empirical work. Koffka (1935, p.153) referred to Gottschaldt (1926), who presented his subjects with simple line patterns "A" for one second each,  such as Figure 4-1A. The subjects had to memorize these figures. Afterwards, they were presented with another set of pictures "B", including Fig. 4-1B, for two seconds. They were asked to describe the pictures "B", mentioning if anything particular struck them about these pictures. Since each B-picture geometrically contained an A-picture, Koffka (1935, p.156) argued:

               Figure 4-1. The B figure contains the A figure

In the case of Fig. 4-1, three subjects were shown the A figure three times and eight subjects 520 times, however, no significant difference between the two groups, in regard to any influence of the memorized A configuration on the B perception, was found.

In this Part, it will be shown that this result does not disprove the empiristic hypothesis, as the gestaltists used the phenomenological memory concept, but the ETVG uses the functionological memory concept. According to Schacter (1987, p.501),

this is the case in the Gottschaldt experiments. But

 In both types of memory, the author assumes, that the experience of a certain configuration, for example, is memorized and can thus later lead consciously, or unconsciously, to the experience of the same configuration.  In this Part, a gestalt factor is assumed to be a memory content. It is not conscious entities ("experiences"), however, that are memorized and lead unconsiously to the same experience. Instead, it is unconscious entities ("relationships between functions") that are unconsciously, unintentionally, memorized and do not actualize any previous experience, but create the first experience of these previously unexperienced relationships. Schacter's concept of memory, although he distinguishs between implicit and explicit memory, is a phenomenological concept, as it requires experiences to be memorized. Thus it is but a reproductive memory, which always produces already known qualities, or configurations. The ETVG concept of implicit (not explicit) memory is a functionological concept, as it does not assume experienced (phenomenal) entities, but unexperienced (functional) entities, to be memorized. Thus it is a creative memory producing only experiences that have never been before.

The concept of implicit learning and memory, most similar to the ETVG concept, seems to be that of Reber (1989), as he does not seem to assume experiences to be memorized. A few citations may illustrate this appraisal:

The aim of this Part is to solve this problem.

Accidentally, Reber (1989) distinguishes between "primitive" and "sophisticated" unconscious:

When compared with the ETVG, the "primitive" unconscious refers to the psychic consciousness (PC), and the "sophisticated" unconscious refers to the mental consciousness (MC) (Fig.10-7). A definition of "implicit learning" which is in agreement to the ETVG, is also the following (Kleine-Horst (1994b, p.41): Implicit learning is the process of ontogenetically early beginning stable memorization, hierarchically constructed, of non-experienced, and thus not recollect- ed, abstract relationships and relation-relationships between sensory percepts, and is the precondition for the development of explicit learning. See also Chapter III.

3. The hierarchical process of memorization

The memorization of the relationships that constitute this structure does not occur simultaneously, but progressively, due to the hierarchical nature of the process. In 1961, I already presumed the visual perception system of a newborn to be a "tabula rasa", which means that the newborn is not immediately capable of gestalt perception. I also presumed that the most frequent relationships between that what is perceived must be memorized the most.

Therefore when a baby is often presented with the perceptual entities A and A1 that are in a (not perceived) relationship B to each other, then the "relationship B between A and A1" will be memorized. After this relationship has been memorized, the next time that A and A1 reappear, the memory content "A and A1 in the relationship B" will be actualized, this means that B, too, can now be experienced, perceived. Note: before the memory content A-B-A1 is formed, neither A-B-A1, nor even B, could be experienced at all; B is purely a relationship between functions, whatever one understands the term "relationship" to mean. Only after the formation of the memory content A-B-A1 can one experience A-B-A1 or B at all. When B is actualized, a further, previously unknown, gestalt quality, "B", is introduced into the percept.

And thus it continues: the relationship C that exists between B and B1, is memorized, despite it not being "experienceable". With the following perception of B and B1, memory content C will be actualized and then phenomenally occur, in the form of "B and B1 in the relationship C", and thus leads to the previously unknown experiencing of "C" (or "C between B and B1"). Level for level, based on long known memory laws, the function system of the gestalt factors is compiled: as a hierarchical system.

Let us go into more detail.

4. The formation of the gestalt factor Pmldt

On the grounds of the structure of the retinal images of the objects in their surroundings, and based on both the Empiristic Development Hypothesis and the physical factors, one may state the first relationships to be memorized between the functional conditions of perceptual experience, that must be memorized. The factor Pmldt is the lowest gestalt factor, the inputs of which come from the physical factors Zm, Zl, Zd, and Zt. There is no perceptual experience corresponding to these factors, that provide only the possibility to experience something in the aspects of modality (brightness and color), locality, depth, and time. The lowest gestalt factor Pmldt, too, is formed by an implicit learning process, as are all other gestalt factors: all Z-outputs are "collected", so to speak, and this collection is memorized, on the grounds of their pure existence at both the same time and the same location, i.e. according to the "law of contiguity". The actualization of this memory content produces the first visual experience of the baby's life, of "is something", in the aspects of modality, locality, depth, and time, i.e. the experience "here, in front, is now something bright".  

5. The formation of the gestalt factors Dm and Dl

After the baby has implicitly, i.e. unconsciously, functionally, "learned" to perceive "here is something bright", the subjective intensity with which the baby perceives brightness, is in accordance to the strength of the sensory stimuli, and can therefore vary. The baby is not capable of experiencing these differences in intensity, at least not with the factor Pml (the d- and t-aspects are no longer taken into consideration). We know that the perception of "brightness differences" must be determined by another factor. Where does this new factor, which previously was not to be found "in" the baby, come from?

According to our empiristic hypothesis, here we must be dealing with a memory content that has come into being through the memorization of the relationships existing between functional Pml- outputs. But which relationships exist between these outputs? It is the difference in strength of the actualizations of the P-factor, and the difference in the locations at which the actualizations of the P-factor take place.

A difference in the degree of actualization often occurs, as the sensory stimuli impinging onto the retina are of different degrees. These differences in the strength of Pml-actualization, i.e. of bright- ness (and color), are memorized, i.e. become a memory content Dm, which, as soon as it is formed, is actualized as the gestalt factor Dm by the following actualizations of Pmls, and thus introduces the gestalt quality "brightness difference" (Dm) into the already existent Pml-percept.

One can argue that there are also equal degrees of actualizations of Pml, which must also be memorized. Of course, there are, however, one could ask: are there more large or more small brightnesses to be memorized? We can answer this question quite simply: since the absolute scale of a dimension is transformed to a relative scale (see Part 2), each two values of Dm are of equal distance from the zero on the relative Dm-scale, however, in opposite directions. This means, that only contrary relationships are memorized, one of which concerns the one area, and the other relationship the other area, adjacent to the previous one.  As these pairs of contrary relationships (here: Dm+ and Dm-) logically occur with equal frequency, they are memorized in equal magnitude. As only pairs of one large and one small brightness difference can be actualized, a surplus Dm+, which does not have a Dm- partner in its immediate vicinity, cannot lead to an experience of brightness difference. This is true also for a Dm- without an adjacent Dm+ partner. (This relationship is, of course, valid also for other gestalt factors - see Part 3:  "Filling-in" per "command from above").

Before the start of the Dm-memorization process, the quality Dm is not experienceable, because the factor Dm does not exist: it will first be formed by the memorizing of those relationships, that later will be the gestalt stimulus for the actualization of the factor (i.e. memory content) "Dm". The memorized relationships are thus pure functional relationships. This is the real reason for assuming these memory contents to be formed unconsciously, beyond the phenomenal sphere, i.e. functionally, or "implicitly".                                   

After the implicit memory content Dm has been formed, the baby can experience the intensity of difference in brightness, and can also detect "large" and "small" brightness differences. The baby could previously detect the brightnesses itself, but now that the factor Dm has been formed it can also perceive the differences in brightness. According to the law of contiguity, we memorize what is found in spatial and temporal proximity to each other. This is here also the case: "large" and "small" brightness differences find themselves in direct spatial proximity at almost the same point in time.

The actualization of the P-factor does lead not only to the experiencing of "something is bright", but also to the experiencing that something bright is "here". Now, the "heres", which are not "experienceable" but only functionally given locations of brightness experiences, are different: there are brightnesses both "here" and "there". Therefore we must assume, that also the differences in location where the brightnesses occur are memorized, and also at "large" and "small" antagonistic values. With the help of the memory content Dl, it is possible to detect brightnesses as being "close together" or "far apart". This theoretically required memory content Dl is already known to us as the gestalt factor Dl, with its positive function Dl+ and its negative function Dl-.

6. The formation of the gestalt factor Gml 

Memory contents that have only just been formed, and whose formation or firmation is still continuing, are "weak". Therefore, more of the gestalt stimuli Dm and Dl are necessary to actualize these still weak memory contents (gestalt factors), Dm and Dl, than will be necessary later, when the memory contents are consolidated.

The kind of strong gestalt stimuli, needed for the actualization of "large" and "small" brightness differences, are provided on the optical borders of the "objects in their surroundings" that are being projected onto the retina. The here provided relationships between Pml-Dm-Dl percepts are memorized and thus lead to a further memory content. So which relationships are here to be found in abundance? It is the relationships we already know as the gestalt stimuli, in the form of a large quantitative ratio of brightness difference to the location difference across which the brightness difference extends (Gml+). Next to which, gestalt stimuli are in the relationship of a small ratio of brightness difference to the location difference across which the brightness difference extends (Gml-). Both relationships and their interrelationships (Gml-/ Gml+/Gml-) are memorized and lead to the appropriate gestalt factor, whose two antagonistic gestalt functions are capable of detecting the relationship of inhomogeneities flanked by homogeneities in the patterns of the following stimuli. Because the Gml+ and Gml- relationships encompass all relationships to be memorized between Dm and Dl, there are no remaining relationships to be memorized. If there are "surplus" Gml+ (or Gm-) relationships to be found between the actualized Dm- and Dm+ functions, they do not come into question for either the memorization or the actualization of the (memorized) Gml relationships (Gml functions), because they lack a complementary partner, within a relatively small retinal Gml double area. As an example: within the "moon", many of the stimulus homogeneities are lacking stimulus inhomogenous partners, in a small retinal area, via which the phenomenally effective Gml-triad could be realized.

7. The formation of the gestalt factors Ll and Fl

Now it is easy to determine the relationship between a number of perceptual triads Gml- / Gml+ / Gml- that will be memorized: it is that which can be termed as "row of Gml+s" and "cluster of Gml-s", as this stimulus pattern, that results from the projection of objects in their surroundings, is such that at the optical projection edges of the objects Gml+ and Gml- will be perceived, and that the Gml+ gestalt qualities are lined-up, next to which, in turn, the perceived Gml-s are arranged on both sides in clusters. The row-arrangement of Gml+s, along with the cluster-arrangement of Gml-s, is an experience made by the visual system "100 to 1000 times per second". This experience does not, however, take place consciously, i.e. in the phenomenal sphere, but sub-consciously, i.e. in the functional sphere. As long as the relationships between Gml+ and Gml- are not memory contents (and therefore not gestalt functions) the rows and clusters will not subjectively be perceived. Only after these functional row and cluster relationships between the Gml+ and the Gml- experiences have been implicitly memorized, can the row relationships themselves be (subjectively, phenomenally) experien- ced (as a "borderline"), and can the cluster relationships themselves be experienced (as a "field").

What is valid for the relationships already described, is valid also for the Fl+ (closedness) and the Fl- (openess) relationships that occur purely functionally between the Ll- / Ll+ / Ll- percepts; they will be memorized and after the next following Ll-perception will stand in the relationship Fl+/Fl-; they lead, when actualized, to the quality "closedness/openness", and, along with their lower-level gestalt factors Pml to Ll, to the percept "figure in its outfield".

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

"On the structure and function of memory"

Back to table of contents