Mind, Brain, and Machine: Object Recognition

E-ISSN： 1943-2860|19|4|555-577

ISSN： 0090-3604

Source： American Academy of Psychoanalysis. Journal, Vol.19, Iss.4, 1991-12, pp. : 555-577

Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.

Previous Menu Next

Abstract

The synthesis of psychodynamic theory and practice with neurosciences and neuropsychiatry, Freud's original intent, may be aided by consideration of developments in machine intelligence. In the domain of object recognition and representation, such cross-talk arises from consideration of the subdivisions in the way the brain sees that have anatomic bases. The brain subdivides recognition tasks by hemisphere, and this subdivision resembles the platonic distinction between basic forms recognized by the left hemisphere and the particular spatial dimensions recognized by the right. Machines recognize objects by means of the mathematical analysis of discontinuities at boundaries, a simplified recognition strategy that perceives cartoon-like entities and bears comparison with the more primitive object recognition by borderline patients, who perhaps are aptly named in this respect. Computational neural networks of the hierarchical feedforward type with back propagation can be made to create internal representations of the objects they recognize and generalize features in ways that model psychopathological states in aging, dyslexia, and schizophrenia. Computer representation is also platonic and architectonic in that shapes are built up in wire form from formulaic solid geometry, with the later addition of perspective, overlapping, light source, shadow, translucency, and texture. Identification, especially of faces, and the common and underappreciated syndromes of misidentification can be analyzed from the viewpoint of neuropathological malfunction and provide insight into the embeddedness of transference in neuromental function.