The cognitive function and framework of the functional hierarchy (Paper V)

Framework for the functional hierarchy of cognitive functions. Records from 3D Brain and multivocal literature were used to collect evidence that supports evidence-based guidelines for determining what types of computing functions can be considered cognitive. The constructed framework underpins human cognitive functions and categorises cognitive computing functions into a functional hierarchy through which the functional similarities between cognitive service and human cognitive functions are presented (Figure 11). The framework construction is focused on the functions of neural systems associated with brain structures. The 3D

66

Brain mobile application illustrates and explains the brain structures with associated cognitive functions. First, information regarding the brain structures and associated cognitive functions were collected and drafted in a graph. Second, the lobes of the brain and their structures were illustrated.

Then, the associated cognitive functions were linked to the structures of the lobes. Third, links were constructed between structures by following the description information. In the analysis of the contents, it was found that the functions form the chain of processes for performing a task, for example visual perception. It was also found that some processes, such as language processes, need further description. Therefore, the structures and associated functions of the brain presentation were supplemented.

Literature searches were used to search for the supplemented information;

these supplements are shown with descriptions of processes with references and/or cited statements.

Figure 11.Framework for the functional hierarchy of cognitive functions (adapted from Paper V).

Mapping human cognitive functions onto groups of cognitive functions. Construction of groups of cognitive functions was performed to map human cognitive functions onto groups of cognitive functions. Human cognition involves multi-functional cooperation (Figure 2). Construction of the groups of cognitive functions and analysis of the contents were performed to assess

67 data and construct categories. The brain functions were grouped according to nine main processes. The groups were obtained from two sub-assemblies: the intentional functions of the brain, and the paths of the transition of the sensory stimuli to the cognitive processes. If the associated cognitive function of brain structure occurred in more than one process, it is listed and mapped in each participant process because human cognitive function results from multi-functional cooperation (Paper V, Figure 2).

Abstraction to cognitive functions functional hierarchy. This addresses the similarities between the functions in cognitive computing and human cognitive functions and summarises the differences and similarities between the functions of the cognitive services and applications that are not defined as cognitive (NDC; Figure 12). These groups of cognitive functions form a functional interactive hierarchy. The associated cognitive functions are grouped adjacent to their respective sensory stimuli as visual, auditory, motor, sensation, and homeostasis functions. Further, language, emotion, and behaviour-related functions of multi-processes are presented. Memory-related functions are collected into memory functions, and finally, higher complex executive cognitive functions are included cognitive functions (Figure 12).

68

Figure 12. Hierarchy of cognitive functions (adapted from Paper V, Figure 4).

The functionality of some cognitive service examples with online demonstrations can be evaluated by experiment to characterise the cognitive computing functions. The McKinsey Global Institute analysis (Exhibit 4, p. 37), for example, identified the social, cognitive, and physical patterns of capabilities which are often required to support many activities.

Hence, chosen functionality examples of cognitive services and applications that are NDC are examples of these capabilities. Their functionality was described and mapped to the brain functions, and their similarities are presented. The framework was used to classify and hierarchise cognitive service functions. For comparison, NDC application functionalities were mapped to brain function, and their similarities are presented. The results are summarised based on content analysis, comparison, and quantification, which presents the differences and similarities between cognitive computing and NDC function. Human cognitive function also results from multi-functional cooperation, for example the functionality of language

69 processes. Similarly, cognitive services contain many functions that produce cognitive outputs: the similarities between human cognitive functions and the functions in cognitive computing. In comparison, it has been found that human cognitive functions are not compatible with cognitive service functions because the extent of functionality is not the same. The functional similarities are presented through the hierarchy between cognitive services and human cognitive functions to illustrate what kind of functions are cognitive in computing. Further, cognitive computing functions produce cognition results with language, emotional, and behavioural interpretation capabilities. The functions go beyond human abilities, for example using big data’s retrieval quantity, speed, and variety, and they combine data from different sources. However, the extent of functionalities, such as human visual recognition regarding images that require imagination to interpret, is beyond the cognitive services. Similarities can be found in visual, auditory, motor, sensation, and homeostasis functions, language and emotion, behaviour functions, memory functions, and cognitive functions (meaning that they can be found in all groups of human cognitive functions). The differences and similarities between the functions of the cognitive services and NDC applications are summarised. NDCs represent applications with rules and pre-determined processes for producing results. Comparisons made through the hierarchy showed that both types of solutions produce results in cognitive functions and that the results of computational functions are cognitive. In general, the computational functions produced cognitive function results since they were implemented to serve a specific purpose. In other words, they were deployed to serve human-origin cognitive needs, for example calculation. Further, differences were found in the cognitive functions of functional hierarchy categories in the following hierarchies:

language, emotional and behavioural functions, memory functions (e.g., learning), and cognitive functions (e.g., interpretation of sensations).

Summarising the above three constructions, 137 human cognitive functions were studied and compared to cognitive services. The IBM Tone Analyzer functionalities were similar to 65 human cognitive functions. The IBM Visual Recognition functionalities were similar to 27 human cognitive

70

functions. The Microsoft Speaker Recognition functionalities were similar to 45 human cognitive functions.