Perception theory

Local and general theory of visual perception

"... there is a big difference between those who view everything from the point of view of a single central visual perception, which is a more or less related or clearly expressed system, on the basis of which people understand, think and feel, that is, those who are guided by one , universal and organizing principle, and those who focus on many aspects of the phenomena, often not related to each other and even contradictory and, if connected, then only by some nodes of a de facto psychological or physiological nature, o not determined by any ethical or aesthetic principles ... "

I. Berlin (1953)

When we enter the room, it seems to us that we see the whole picture before our eyes at one glance. In reality, visual perception is a long process. Everything takes time: both to review the details and gather the necessary information about them, and to make assumptions, to check them and come to certain conclusions, and to evaluate the data obtained taking into account the objectives pursued, its own knowledge and expectations. Nevertheless, all this happens so quickly and naturally, which involuntarily causes surprise and requires explanation.

Some experts consider unsatisfactory those theories in which an attempt is made to interpret the phenomenon of visual perception from the position of discrete, sequential, symbolic processes. It seems to them that although machine programs written on the basis of these theories can indeed demonstrate the effect of "vision", they are unacceptable for living beings because of the roughness of perception and low speed. However, the commonly proposed alternative, referring to the extreme case of holism or the idealistic "philosophy of integrity" (see VG Afanasyev, 1964), cannot be technically implemented. Below I will argue that the following one-by-one character processes can explain the essence of the phenomena that seem instant and complete to us and that take place in the analysis of scenes.

Some theorists who adhere to the early Gestalt-psychological concepts (for details, see MG Yaroshevskn, 1976), tried to explain the phenomenon of visual perception with the help of the electric field in the human brain, but in the 30s of our century, the inconsistency of such views became clear . Their modern followers, including adherents of the idea of ​​integral transformations, the use of holograms and interference phenomena, have also failed to achieve much. And yet, despite these setbacks, most experts still believe that the required perception speed can only be achieved through a global, parallel process like the wave process.

The frame theory presented in this paper addresses all the main issues raised by gestalt psychologists. Common to both approaches is the desire to uncover the essence of the processes underlying the correlation of sensory data either with general concepts or with their parts, but the methods proposed to substantiate the advanced positions are radically different from each other. Gestalt psychologists sought to prove that these processes are mainly based on the interaction of a small number of universal and very effective procedures, but failures with their effective release showed the limitations of this idea. In frame theory, emphasis is placed on the multiplicity of interactions between sensory data and the vast network of symbolic information formed in the process of learning. And although these interactions should, ultimately, be based on one or another set of general principles, in our theory, the study of thought processes is separated from questions of the origin and development of AI systems.