| Both sides previous revision
Previous revision
|
|
research_on_wba [2016/01/03 21:21]
n.arakawa removed |
— (current) |
| ==== Researches on WBA ==== | |
| Some researches conform to the specification of the [[WBA Product]] and others may be more vaguely related to the [[http://wba-initiative.org/en/wba/|the WBAI approach]].\\ | |
| While the specification of the [[WBA Product]] is rather abstract, this page refers to concrete brain organs to give an 'image' of WBA.\\ A WBA would have to incorporate at least the first three components (the perceptual system, the hippocampus, and the pre-frontal-cortex-basal-ganglia-thalamus loop). | |
| |
| Note that this page is under construction. | |
| |
| === Two-Stream Perceptual System === | |
| It is hypothesized that the visual and auditory systems in the brain are separated into [[https://en.wikipedia.org/wiki/Two-streams_hypothesis|ventral and dorsal streams]].\\ The feature-place separation by the two streams would be related to [[https://en.wikipedia.org/wiki/Binding_problem|the binding problem]] (at least the binding of information on perceptual features and their locations).\\ | |
| At least the ventral stream consists of a cascade of cortical regions, forming a 'deep' network. A model also has to take into account the fact that the cascade is richly bi-directional with afferent and efferent connections. | |
| === Hippocampus === | |
| The hippocampus, consisting of several sub-regions, is supposed to be responsible to the following cognitive functions among others. | |
| * Transferring mid-term memory into long-term memory | |
| * Navigation (at least in rodents; see [[http://www.scholarpedia.org/article/Grid_cells|the grid cells and place cells]]) | |
| As the place cells respond to the places the subject has visited, it is supposed to be related to long-term memory.\\ | |
| The circuitry of the hippocampus has been well studied. | |
| === Pre-frontal Cortex, Basal Ganglia and Thalamus === | |
| These organs, forming a circuit (loop), are supposed to be involved in (action) planning and execution.\\ | |
| The basal ganglia are hypothesized to use reinforcement learning principles.\\ | |
| For a computational model, see\\ | |
| O'Reilly et al.: [[http://psych.colorado.edu/~oreilly/papers/OReillyFrank06_pbwm.pdf|Making Working Memory Work: A Computational Model of Learning in the Frontal Cortex and Basal Ganglia]], Neural Computation, 18, pp.283-328 (2006) | |
| === Amygdala === | |
| The amygdala is supposed to be related to affect.\\ | |
| See [[http://www.scholarpedia.org/article/Amygdala|Amygdala@Scholarpedia]] for detailed explanation. | |
| === 'Language Areas' === | |
| If you make a human-level WBA model, you definitely have to take the 'language areas' into account.\\ | |
| [[https://en.wikipedia.org/wiki/Wernicke%27s_area|Wernicke's area]] and [[https://en.wikipedia.org/wiki/Broca%27s_area|Broca's area]] are two well known language areas, where the former is hypothesized to be involved in language comprehension and the latter speech production.\\ | |
| For a computational model, see\\ | |
| P. F. Dominey: [[http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3733003/|Recurrent temporal networks and language acquisition—from corticostriatal neurophysiology to reservoir computing]], Frontiers in Psychology, 4: 500 (2013). | |
| ==== References ==== | |
| * [[https://grey.colorado.edu/CompCogNeuro/index.php/CCNBook/Main|The Computational Cognitive Neuroscience Book]] | |
| * [[http://blog.agi.io/2015/12/how-to-build-general-intelligence.html|How to build a General Intelligence: Circuits and Pathways]] | |
| * [[http://www.reservoir-computing.org|Reservoir-Computing.org]] | |