6.4: Conclusion and Implications

Both the cognitive and the neuropsychological approaches to studying the visual system and object recognition have revealed unique solutions and more interesting questions. Yet, there is much still to be learned and integrated across these domains. Indeed, computational cognitive models have been exceedingly informative, particularly in regard to application in computer vision. What can be gleaned from integrating these approaches to object recognition?

The models and approaches mentioned were by no means exhaustive. Instead, the purpose of the present chapter was to investigate what perceptual and object recognition evidence exists within both neurophysiological and cognitive studies that provides a foundation for conceptualizing the human brain as a predictive machine. Neurophysiology has demonstrated where information is entering, how it proceeds, and demarcated what cells and areas are doing distinct work. However, a number of these models do not incorporate theory on dynamic, integrated processing, instead focusing on isolated units. Cognitive theories diverge on this point. The center around extensive information processing formulations. Though some still consider modular accounts of cognition, the orientation toward process is valuable and something to be incorporated in future research.

Informed by neurophysiological findings, we see advances in abnormal mind perception. From schizophrenia (Butler, Silverstein, & Dakin, 2008) to autism spectrum disorder (Dakin & Frith, 2005; Grice et al., 2001), the etiological underpinnings of nonconforming minds may be better understood by defining specializations of specific cortical areas in the brain. Studies that are concerned with representations and information processing have been particularly helpful in the technological domain as computers work like this. For example, major advances in computer vision have been spearheaded by cognitive computational models (Brown, 1985; Ullman et al., 2016). Recently, attempts at computer vision have shifted to convolution networking to better match the complexity and accuracy achieved by the human brain (Simonyan & Zisserman, 2014). Again, there are clear applications of this work and the theories that emerge through trials are potentially informative for a number of other phenomena. Namely, perception is one link to the consciousness. Understanding how prediction and sensory information are integrated in our brain can provide a meaningful step toward solving problems of consciousness.

The biases that predictions and expectations produce exist elsewhere (such as visual attention) and have profound effects on downstream cognitive processes like judgments and behavior. For instance, evidence clearly shows that biases influence important social customs like eye-witness reporting (MacLeod, 2002; Storbeck & Clore, 2005). The consequences of these decisions are clear and understanding how emotion regulates and biases affective feelings and associations may contribute to understanding how to recalibrate social systems where a great deal of the ultimate decision depend on individual accounts.