Visual perception

Does visual perception have a symbolic form?

Is it really possible to assume that human perception of three-dimensional objects is so fragmentary and atomistic that it can be represented in terms of the relationship between elements of two-dimensional images? Let us immediately separate these two questions from each other: are visual images symbolic and are they based on two-dimensional structures? The first question is of particular importance; There is no doubt that at some level visual perception proceeds mainly in symbolic form. Disagreements can be between those who, on the one hand, adhere to several naive concepts and consider their acceptance as either picture-like or proceeding on the basis of operating with imaginary geometric bodies, and those who, on the other hand, are based on experimental data (see W .Piazhe, B.Inelder, 1956, and others), proves that many possible limitations arising from symbolic representations do indeed exist.

So, we know that children in their work, especially in graphics (however, this also applies to adults) use a set of very limited symbolic ingredients (see, for example, E. Gombrich (1959)). Prospects and shielding are usually presented not as they really are, but with the help of certain conventions. The metric ratios are highly distorted; complex shapes are depicted using special characters that are not used to represent the most significant! signs. Representatives of the "naive" point of view usually do not recognize such tricks and hold the opinion that people really "see and manipulate images like these pictures" in such a way that this cannot be explained with the help of discrete descriptions.

As for the second question (whether the images are two- or three-dimensional), its symbolic descriptions do not exist at the level, because the very concept of measurement becomes inappropriate here. Each type of symbolic description of an object serves one purpose well, and badly to another. If "left-from", "right-from" and "is-above" relations between elements of a certain structure are specified and they are represented as markers defined on pairs of terminals, then during certain manipulations with the object its description is made on this basis will be sufficient to predict the location of its individual elements. The task is facilitated by the fact that if, for example, you rotate a cube without changing its orientation in space (without changing the face with which it touches the table), then certain properties of these relations will be invariant to such movements. Most items usually have their upper and lower parts. However, if you put the cube on the side, predictions based on the same descriptions will be much harder to do: people have great difficulty tracking the edges of a six-color cube (i.e. a cube, each face of which is colored in a different color), if make them mentally rotate it.

If for the same purposes we use more “characteristic” relations, such as “next-to” or “to be-opposite-to,” then similar descriptions of images will be less sensitive to possible turns of objects. In the works of P. Winston (1970, 1971, 1972) we see how systematic substitutions of relations (for example, "left" instead of "back" or "right" instead of "front") can be used to simulate the rotation of objects.

W. Hogarth condemned those artists who had devoted too little time to perfecting their ideas about the objects around them. (William Hogarth (1697-1764), an outstanding English painter, graphic artist and art theorist, published his famous theoretical treatise Beauty Analysis in 1753). He advised those who seek to get the right ideas about distances, relationships and differences between some significant points and lines belonging, in the worst case, even to the most asymmetrical figures, to gradually develop the ability to extract them from their memory, because it can greatly help to the one who constantly invents something or draws from memory and contributes to the accurate full-scale reproduction of objects.

Thus, the intentional training of memory in questions of systematization of relations between points lying on opposite surfaces of bodies is, according to W. Hogarth, the key to understanding the invariant relations between visible and invisible parts of images; they can give a person enough information to imagine himself inside some object or mentally find himself in another, almost inaccessible point of observation. From this we can conclude that W. Hogarth rejected the “naive” concepts in the theory of perception of images.

Some people believe that spatial problems are solved with the help of an analogue of a three-dimensional structure that is somehow stored in memory. If, however, someone could recreate such a model, then for the “intellectual eye” most of the traditional problems that relate to the real eye would remain, and, moreover, a very difficult new task would appear: creation (on basis of two-dimensional structures) image of some hypothetically imaginary object.

Although these arguments, as it may seem, indicate the feasibility of using two-dimensional images for aggregation and pattern recognition, they cannot be considered satisfactory for the tasks of planning and performing manipulative operations. A more natural look is another way of presenting information in the same symbolic form, but on the basis of basic geometric forms. Thus, the handset can be described using two truncated spherical bodies connected by a curved rectangular rod. In the next paragraph we will discuss the issue of sharing two or more methods, qualitatively different from each other, for representing the same object.