Pages:     | 1 |   ...   | 52 | 53 || 55 | 56 |   ...   | 58 |

It is therefore necessary to understand, In addition to these main components, a through a study of a variety of problems and BICA needs to specify elements and rules of solutions, ranging from navigation to dynamics. One essential dynamic law is the emotional intelligence and goal generation, cognitive cycle, that includes at least three why existing implementations do not match elements: perception, cognition, and action.

the human level, and how should they be As an agent, a BICA is embedded in some implemented in order to match the goal environment: virtual or physical, where it can requirement. One negative example that can perceive sensory input and perform actions.

be named here is the so-called virtual person (http://ta-vision.blogspot.com/), claimed to be Discussion of Challenges based on models of artificial emotional intelligence, yet is not capable of producing There are many steps leading us from this believable behavior, and is not very different generic BICA design up to the solution of the from the old good ELIZA and ALICE [23].

BICA Challenge. The necessary ones are Nevertheless, based on the present state of the motivated by the intended role of BICA in art, a solution to the BICA Challenge can be future applications [4]. Some of these steps are expected in a near future.

bottlenecks, or the key challenges, that But, is it the right thing for our society need to be solved before the ultimate goal can to find and implement this solution I am be reached. On the greater roadmap (a arguing that it is, at least, for the following simplified version of which is represented in reason. Modern computers require continuous Figure 3), they appear as parts of the general human assistance beyond their programmed BICA Challenge. The critical capabilities behavior. As a result, many people spend half include: the feeling of presence, metacognitive of their active life time sitting at a computer.

capabilities and believable behavior, that in When the virtual human-level team member turn are based on the Theory-of-Mind will become available, the situation may capabilities, self-regulation, emotional change. Many routine human jobs could be intelligence, teleological capabilities, imagery, performed autonomously by computers sensemaking, etc. (Figure 3).

themselves, with minimal human instruction and guidance, that can be limited to a higher, XVI 8. Gluck, K. A., and Pew, R. W. (Eds.). (2005).

metacognitive level. As a result, humans will Modeling Human Behavior with Integrated Cognitive regain their freedom Architectures: Comparison, Evaluation, and Validation.

Mahwah, NJ: Erlbaum.

Conclusions 9. Gray, W. D. (Ed.) (2007). Integrated Models of Cognitive Systems. Series on Cognitive Models and Architectures. Oxford, UK: Oxford University Press.

In conclusion, it is argued here that BICA 10. Laird, J.E., Rosenbloom, P.S., and Newell, A.

is a new, powerful modeling approach that (1986). Universal Subgoaling and Chunking: The will allow us to make progress toward Automatic Generation and Learning of Goal building a computational equivalent of the Hierarchies. Boston: Kluwer.

human mind, combining symbolic and 11. Laird, J.E., Newell, A., and Rosenbloom, P.S., (1987). SOAR: An architecture for general intelligence.

neuromorphic approaches. A solution to the Artificial Intelligence 33: 1-64.

BICA Challenge can be expected in the near 12. J. E. Laird (2008). Extending the Soar cognitive future based on this approach. The problem architecture. In P. Wang, B. Goertzel and S. Franklin, requires joined efforts of experts from many eds. Artificial General Intelligence 2008: Proceedings of fields of science. The impact of the solution the First AGI Conference, pp. 224-235. Amsterdam, The Netherlands: IOS Press.

hardly can be overestimated.

13. Anderson, J. R. and Lebiere, C. (1998). The Atomic Components of Thought. Mahwah: Lawrence References Erlbaum Associates.

14. Anderson, J. R. (2007). How Can the Human 1. McCarthy, J., Minsky, M.L., Rochester, N., & Mind Occur in the Physical Universe New York:

Shannon, C.E. (1955/2000). A proposal for the Oxford University Press.

Dartmouth summer research project on artificial 15. Lenat, D. B. (1995). CYC: a large-scale intelligence. In Chrisley, R., & Begeer, S. (Eds.).

investment in knowledge infrastructure.

Artificial Intelligence: Critical Concepts. Vol. 2, pp. 44Communications of the ACM 38 (11): 32-38.

53. London: Routledge.

16. McCord, M.C., Murdock, J.W., and Boguraev, 2. Newell, A. (1990). Unified theories of cognition.

B.K. (2012). Deep parsing in Watson. IBM Journal of Cambridge, MA: Harward University Press.

Research and Development, 56 (3-4). DOI:

3. Chella, A., Lebiere, C., Noelle, D.C., & 10.1147/JRD.2012.2185409.

Samsonovich, A. V. (2011). In: Samsonovich, A. V., & 17. Nadel, L. and Oliver, H. (2011). Update on Johannsdottir, K. R. (Eds.). Biologically Inspired memory systems and processes.

Cognitive Architectures 2011: Proceedings of the Neuropsychopharmacology 36: 251-273.

Second Annual Meeting of the BICA Society. Frontiers 18. Samsonovich, A. V. and Ascoli, G. A. (2010).

in Artificial Intelligence and Applications, vol. 233, pp.

Principal Semantic Components of Language and the 453-460. Amsterdam, The Netherlands: IOS Press.

Measurement of Meaning. PLoS ONE 5 (6): e10921.14. Samsonovich, A. V. (2012). On the roadmap for e10921.17.

the BICA Challenge. Biologically Inspired Cognitive 19. Tulving, E. (1983). Elements of Episodic Memory.

Architectures, 1 (in press).

New York: Clarendon Press.

5. SIGArt, (1991). Special section on integrated 20. Zimmer, H.D., Cohen, R.L.,Guynn, M.J., cognitive architectures. Sigart Bulletin, 2(4).

Engelkamp, J., Kormi-Nouri, R., & Foley, M.A. (Eds.).

6. Pew, R. W., and Mavor, A. S. (Eds.). (1998).

(2001) Memory for Action: A Distinct Form of Episodic Modeling Human and Organizational Behavior:

Memory Oxford, UK: Oxford University Press.

Application to Military Simulations. Washington, DC:

21. Doeller, C. F., Barry, C., and Burgess, N. (2010).

National Academy Press.

Evidence for grid cells in a human memory network.


Nature, 463 (7281): 657-661.

7. Ritter, F. E., Shadbolt, N. R., Elliman, D., Young, 22. Samsonovich, A. V., De Jong, K. A., and R. M., Gobet, F., and Baxter, G. D. (2003). Techniques Kitsantas, A. (2009). The mental state formalism of for Modeling Human Performance in Synthetic GMU-BICA. International Journal of Machine Environments: A Supplementary Review. WrightConsciousness 1 (1): 111-130.

Patterson Air Force Base, OH: Human Systems 23. McCorduck, P. Machines Who Think. W. H.

Information Analysis Center (HSIAC).

Freeman, 1979.

4- ۻ SEROTONIN-EVOKED REORGANIZATION OF THE BRAIN STEM RESPIRITORY NETWORK: INSIGHTS FROM COMPUTATIONAL MODELING N. Shevtsova1, A. Bischoff2,3, Y. Molkov1,4, T. Manzke2,3, J. Smith5, D. Richter2,3, I. RybakDepartment of Neurobiology and Anatomy, Drexel University College of Medicine, Philadelphia, PA, USA Department of Neuro- and Sensory Physiology, University of Gttingen, Gttingen, Germany DFG Research Center of Molecular Physiology of the Brain, Gttingen, Germany Department of Mathematical Sciences, Indiana University-Purdue University, Indianapolis, IN, USA Cellular and Systems Neurobiology Section, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD, USA Natalia.Shevtsova@drexelmed.edu Activation of the serotonin (5-HT) 1A receptors was Model description recently found to have therapeutic effect for breathing recovery after opioid-evoked suppression. This recovery Our previous computational models of the was postulated to originate from an enhanced brainstem respiratory network reproduced glycinergic synaptic inhibition associated with the many experimental phenomena [5, 7, 9]. These glycinergic 3 receptors in some inhibitory neurons models differed in details incorporated to leading to disinhibition and reactivation of their targets.

A computational model of the brainstem respiratory simulate particular aspects of neural control of network was developed to evaluate this proposal and breathing; however, all of these previous investigate the neural mechanisms involved models had the same core neural circuitry consisting of four populations of respiratory Introduction neurons located in the Btzinger and preBrainstem respiratory neurons express the Btzinger complexes (pre-BtC and BtC, see glycine 3 receptor (Gly3R) that is a target of Fig. 1A). The core network included (i) an several 5-HT receptor (5-HT-1AR) agonists. excitatory pre-inspiratory inspiratory (pre-II) Application of the 5-HT-1AR agonist population of neurons and (ii) three mutually 8-OH-DPAT was shown to induce (1) an interacting inhibitory populations: postaugmentation of postsynaptic inhibition of inspiratory, post-I, augmenting-expiratory, neurons expressing Gly3R, and (2) a dose- aug-E, and an early inspiratory, early-I(1) dependent hyperpolarization of respiratory populations. These models also included an neurons via increased conductances of 5HT- output compartment (rVRG) that contained a activated potassium (K+) leak channels [4]. excitatory ramp-inspiratory neural population, Activation of 5-HT-1AR was recently found to ramp-I, projecting to phrenic motoneurons and protect and/or restore breathing during opioid an inhibitory early-I(2) population shaping the pharmacotherapy of pain often accompanied ramp-I firing pattern.

by opioid-induced apnea. This effect seems to However, none of our previous models rely on the enhanced Gly3R-mediated considered different types of inhibitory inputs inhibition of inhibitory neurons causing (glycinergic vs. GABAergic) that could be disinhibition of their target neurons. provided by phenotypically distinct neural To evaluate this proposal and investigate populations. At the same time, the effects of neural mechanisms involved, an established neuromodulators, including 5-HT, and other computational model of the brainstem pharmacological agents (e.g., opioids) are respiratory network [9] was extended by usually dependent on the neuronal phenotype (1) incorporating distinct subpopulations of and specific neurotransmitters and receptors inhibitory neurons (glycinergic and involved in network interactions. To consider GABAergic) and their network synaptic synaptic transmission- and receptor-specific connections within the respiratory network, modulation, we extended our previous model and (2) assigning the 5-HT-1AR-Gly3R (see Fig. 1B), by (i) incorporating distinct complex to some of these inhibitory neuron glycinergic and GABAergic populations and types in the network. their synaptic connections, (ii) assigning the XVI Results Figure 2A1 shows model performance under normal condition. The model generates a typical three-phase respiratory rhythm similar to that generated in the previous model [9]. Under normal conditions the two early-I populations exhibit similar activity profiles, as do both glycinergic and GABAergic aug-E populations. Figure 2A2 shows simulation of application of a low dose of 8-OH-DPAT. In this simulation, early-I(2) neurons, which have Fig. 1. Schematics of (A) the basic model depicting core components of the brain stem respiratory network and Gly3R and receive glycinergic inhibition (B) the extended model proposed in this study. See text from the post-I neurons, are affected in the and legend for details. Modified from [8].

inspiratory phase. As a result, the post-I neurons are released from inhibition and the 5-HT-1AR-Gly3R complex to some neuron onset of their activity shifts to the beginning of types (synaptic weights of glycinergic inputs inspiration (Fig. 2A2). Because aug-E(1) and to these neurons increased during 8-OH-DPAT dec-E neurons have Gly3R, application of application producing 5-HT-1AR-Gly3R8-OH-DPAT augments both the inhibition of dependent potentiating of glycinergic aug-E(1) neurons by the dec-E population and inhibition), and (iii) incorporating 5-HT-1ARthe inhibition of dec-E neurons by the activated K+ leak channels activated by higher aug-E(2) population. This shortens expiration.

doses of 8-OH-DPAT into all neurons. More As a result, the network starts generating a specifically: (i) the population of early-I(1) two-phase rhythm lacking the post-inspiratory neurons in the pre-BtC was divided into a phase and the frequency of oscillation GABAergic, early-I(1), and glycinergic, increases compared to control. The results of a early-I(2), subpopulations; (ii) the post-Idec-E simulated application of a higher dose of population was split into two glycinergic 8-OH-DPAT are shown in Fig. 2A3 when populations: post-I and dec-E with different neurons undergo activation of 5-HT-1ARconnectivity; (iii) the aug-E population was regulated K+ leak channels resulting in a split into separate GABAergic, aug-E(1), and membrane hyperpolarization. As a result, the glycinergic, aug-E(2), subpopulations.

post-I activity transforms into a short late-I The details of the modeling methods can discharge at the end of inspiration (Fig. 2A3), be found in [5, 7-9]. Briefly, all neurons were which causes a further increase in frequency.

Pages:     | 1 |   ...   | 52 | 53 || 55 | 56 |   ...   | 58 |

2011 www.dissers.ru -

, .
, , , , 1-2 .