عنوان: نگرش محاسباتی پراکنده به پردازش داده های سنسوری در قشر سنسوری مغز
Brains and computers work in very different ways, and they are good at very different things. In some areas, such as performing calculus, remembering lists, or following instructions exactly, computers are many orders of magnitude better than brains. But in other areas, such as interacting with the world or reaching reasonable conclusions based on available information, neural systems are many orders of magnitude better (if, indeed, computers can perform these tasks at all). For instance a tiny fruit-fly with just thousands of neurons is able to amazingly handle a complicated sensory-motor processing task to navigate in real-time. This fact necessitates the development of new styles of information-processing with are apparently different from conventional computational architectures that we are deeply used to them.
Despite decades of research, it is not yet fully understood how “cortical processing” occurs above the single cell or small-network level, what the processing consists of, or even how the processed data is represented. This lack of understanding directly impedes the development of technologies with brain-like information processing abilities. We will discuss how this conceptual roadblock can be tackled by developing high-level information processing models which are functionally-distributed and processor-free. By functionally-distributed, we mean that different functionalities are computed in different parts of the system (cortical distributed circuits); by processor-free, we mean that - conceptually - there are no central or even local CPU-like sequential instruction-following elements.
Within the scope of this talk, we will specifically see how a distributed cortical structure can facilitate, and learn the process of sensor fusion and multisensory perception in human - the essential function that we need to interact with our environment and to survive. We will give an overview of underpinning principles of Multisensory Information processing in Human brain , what is going on in the brain during this process , current state of research, challenges and open questions, and we will introduce a distributed probabilistic neural structure for Sensor Fusion during a Multisensory localization task. In this work and through a neural modeling approach, we will demonstrated how such a distributed cortical hierarchy can perform probabilistic causal inference and reproduce human behavior .
- Alais, and D. Burr, (2004). The ventriloquist effect results from near-optimal bimodal integration, Curr. Biol. 14, pp 257–262.
- Rohe, U. Noppeney, (2015) Cortical Hierarchies Perform Bayesian Causal Inference in Multisensory Perception. PLoS Biol 13:e1002073.
- Firouzi, S. Glasauer, and J. Conradt (2015) Causal Bayesian Inference in hierarchical distributed computation in the Cortex, towards a neural model, Bernstein Conference on Computational Neuroscience, Heidelberg, 2015