Quantum materials for brain sciences and artificial intelligence
- PDF / 1,428,736 Bytes
- 7 Pages / 585 x 783 pts Page_size
- 49 Downloads / 183 Views
The brain as an information processor A key function of the brain is to process a multitude of information received through various sensory inputs and make informed decisions. This occurs via two critical components—the neuron and the synapse. A neuron may be simply considered as a cell that transmits nerve impulses in the form of electrical signals, while a synapse is a junction between nerve cells with programmable resistance. The number of neurons, synapses, and their interconnections can vary by several orders of magnitude, depending on the organism. For instance, the mouse has approximately 70 million neurons, while humans have more than 85 billion neurons.1 A small subset of neurons (few to dozens) in a circuit are utilized to process specific types of information or evoke responses to specific stimuli. Common features of neurons across animal species do exist, forming the basis for much neuroscience research. Neurons transmit information via electrical signals known as action potentials.2 These are in the form of electrical spikes (Figure 1). While there exist several types of neurons with varying structure and interconnections, called neural circuit pathways, and physical locations in the brain or elsewhere in the body, they do share similarities in the form of action potentials used to transmit information. The typical voltage spikes are on the order of 100 mV in neuronal signals and time scales are on the order of milliseconds.3 Among the distinct neuron types, there can be a distribution
in the magnitudes and spike time widths. The time scales for signaling in the brain are clearly much slower than transistor switching speeds or interconnect wire delays in state-ofthe-art computer chips. Yet, the brain possesses capabilities unmatched by any computer to date for certain tasks, and this serves as one motivation to design the next generation of machines that can evolve and learn like the brain, in addition to being energy efficient.4 Adaptive and programmable materials are therefore of great interest in this context. Synapses serve to weigh the effective strength of signals transmitted by neurons as they propagate to other neuron branches. Synapses can gate the signals via electrical or chemical mechanisms (Figure 2). Electrical synapses transmit information rapidly (milliseconds), while chemical synapses can be a few orders of magnitude slower (seconds). From an evolutionary perspective, each has their primary function. Electrical synapses can be useful when an organism faces danger, for instance, and has to make rapid decisions to protect its life (survival response), while chemical synapses can massively amplify signals and are considered to be useful in learning.5 Such remarkable diversity in neural components suggests how function may be partitioned across the brain. Perhaps most interestingly, information in the brain is encoded in time. What this means is that the frequency of neuron firing is a key factor in healthy functioning of the brain. Over the past century, measurement of electrical signals
Data Loading...
