Books
What can Artificial Intelligence get from Neuroscience?
In M. Lungarella, J. Bongard, & R. Pfeifer (Eds.), Artificial Intelligence Festschrift: The next 50 years. (pp. 174-185). Berlin: Springer-Verlag.
The human brain is the best example of intelligence known, with unsurpassed ability for complex, real-time interaction with a dynamic world. AI researchers trying to imitate its remarkable functionality will benefit by learning more about neuroscience, and the differences between Natural and Artificial Intelligence. Steps that will allow AI researchers to pursue a more brain-inspired approach to AI are presented. A new approach that bridges AI and neuroscience is described, Embodied Cultured Networks. Hybrids of living neural tissue and robots, called hybrots, allow detailed investigation of neural network mechanisms that may inform future AI. The field of neuroscience will also benefit tremendously from advances in AI, to deal with their massive knowledge bases and help understand Natural Intelligence.
Keywords: Neurobiology, circular causality, embodied cultured networks, animats, multi-electrode arrays, neuromorphic, closed-loop processing, Ramon y Cajal, hybrot.
- 81 Views
-
Like (1)
Closing the Loop: Stimulation Feedback Systems for Embodied MEA Cultures.
Advances in Network Electrophysiology Using Multi-Electrode Arrays, pp. 215-242. Taketani, M. and Baudry, M. New York, Springer
For certain key functions, simple animals use large identified neurons, such as the locust's Giant Motion Detector neuron (LGMD), which integrates visual information and triggers jumping (Gabbiani et al., 1999). By contrast, in the vertebrate nervous system, individual neurons are probably not important; each function is subserved by many nerve cells working in concert. However, we have little understanding of how single-unit activity combines to form the network-level processing that takes in sensory input, stores memories, and controls behavior. Cultured neuronal networks have provided us with much of our present understanding of ion channels, receptor molecules, and synaptic plasticity that may form the basis of learning and memory (Bi and Poo, 1998; Latham et al., 2000; Misgeld et al., 1998; Muller et al., 1992; Ramakers et al., 1991). To study the nervous system in vitro offers many advantages over in vivo approaches. In vitro systems are much more accessible to microscopic imaging and pharmacological manipulation than are intact animals. Recent developments in multi-electrode array (MEA) technology, including those described below, will enable researchers to answer questions not just at the single-neuron level, but at the network level. Most MEA research has involved recording the activity that cultured networks produce spontaneously, via up to 64 extracellular electrodes. While some studies also included electrical stimulation via the substrate electrodes, it was applied to only one or two of them at a time (Connolly et al., 1990; Fromherz and Stett, 1995; Gross et al., 1993; Jimbo and Kawana, 1992; Jimbo et al., 1998; Maeda et al., 1995; Oka et al., 1999; Regehr et al., 1989; Shahaf and Marom, 2001; Stoppini et al., 1997). We propose that in order to substantially advance our understanding of network dynamics, we need high-bandwidth (many neuron) communication in both directions, out of and into the network. This chapter describes technologies that allow recording and stimulation on every electrode of an MEA, and a new closed-loop paradigm that brings in vitro research into the behavioral realm.
- 25 Views
-
Like (1)
Removing some 'A' from AI: Embodied Cultured Networks.
Bakkum, D. J., Shkolnik, A. C., Ben-Ary, G., Gamblen, P., DeMarse, T. B. and Potter, S. M. (2004). Removing some 'A' from AI: Embodied Cultured Networks. Embodied Artificial Intelligence. Iida, F., Pfeifer, R., Steels, L. and Kuniyoshi, Y. New York, Springer. 3139: 130-145.
We embodied networks of cultured biological neurons in simulation and in robotics. This is a new research paradigm to study learning, memory, and information processing in real time: the Neurally-Controlled Animat. Neural activity was subject to detailed electrical and optical observation using multi-electrode arrays and microscopy in order to access the neural correlates of animat behavior. Neurobiology has given inspiration to AI since the advent of the perceptron and consequent artificial neural networks, developed using local properties of individual neurons. We wish to continue this trend by studying the network processing of ensembles of living neurons that lead to higher-level cognition and intelligent behavior.
- 20 Views
-
Like (1)
Two-photon microscopy for 4D imaging of living neurons.
Potter SM (2005) Two-photon microscopy for 4D imaging of living neurons. In: Imaging in Neuroscience and Development: A Laboratory Manual (Yuste R, Konnerth A, eds), pp 59-70: Cold Spring Harbor Laboratory Press.
T wo-photon laser-scanning fluorescence microscopy (Denk et al. 1990) has made it possible to image neurons over 600 pm deep within a living slice or organism, with submicrometer resolution and in three dimensions (3D) for many hours without photodamage (Potter et al. 1996a). With true 4D microscopy (i.e., 3D with time), it is now feasible to capture neural development and synaptic plasticity in the act of happening, eliminating many uncertainties associated with between-animal comparisons of fixed-tissue specimens. By using pulsed IR laser light to excite fluorescent labels (e.g., dyes, fluorescent proteins, or endogenous fluorophores) that are normally excited by visible light, excitation is restricted to the focal plane, greatly reducing photobleaching and phototoxicity (see Chapter 7). The IR illumination is scattered less than visible light, allowing imaging 2-3 times deeper than with standard confocal microscopy. In addition, because the fluorescent signal emanates only from the focus of the scanning IR laser beam within the specimen, no confocal aperture is necessary to remove the out-of-focus signal. This means that two-photon microscopy has an inherently higher signal-to-noise ratio (SNR) compared with confocal microscopy. Excellent references for two-photon microscopy, and for labeling and imaging living specimens, include Denk et al. (1995) and Terasaki and Dailey (1995). In this chapter, I describe two-photon imaging hardware, pointing out potential pitfalls in microscope construction and operation. I also describe technical considerations for successful two-photon microscopy in two experimental systems: (1) 4D imaging of living neurons transplanted to cultured hippocampal slices from neonatal rats and (2) 4D imaging of dendritic spines within acute hippocampal slices from adult rats. Figure 1 shows the components of the imaging setup.
- 28 Views
-
Like (1)
Add Comment