This contribution revisits an earlier discovered observation that the average performance of a pop ulation of neural networks that are evolved to solve one task is improved by lifetime learning on a different task. Two extant, and very different, explanations of this phenomenon are examined- dynamic correlation, and relearning. Experimental results are presented which suggest that neither of these hypotheses can fully explain the phenomenon. A new explanation of the effect is proposed and empirically justified. This explanation is based on the fact that in these, and many other relat ed studies, real-valued neural network outputs are thresholded to provide discrete actions. The effect of such thresholding produces a particular type of fitness landscape in which lifetime learn ing can reduce the deleterious effects of mutation, and therefore increase mean population fitness.
Ackley, D., and Littman, M. (1991). Interaction between learning and evolution. In C.Langton, C. Taylor, J. Farmer and S. Rasmussen (Eds.), Artificial Life II: Proceedings of Santa Fe Conference, (pp.487-509) February 1990. Reading, MA: Addison-Wesley .
2.
Baldwin, J.M. (1896). A new factor in evolution. American Naturalist30, 536-553.
3.
Belew, R.K. (1989). When both individuals and populations search: Adding simple learning to the genetic algorithm. In J.D. Schaffer (Ed.), Proceedings of the Third International Conference on Genetic Algorithms. pp.(34-41). San Matteo, CA: Morgan Kauffmann.
4.
Belew, R.K., and Mitchell, M.(Eds.)(1996). Adaptive individuals in evolving populations. Reading MA: Addison-Wesley.
5.
Cecconi, F., and Parisi, D. (1991). Evolving organisms that can reach for objects . In J. Meyer and S. Wilson (Eds.), Proceedings of 1st International Conference on Simulation of Adaptive Behavior pp.(391-399). Cambridge, MA: MIT Press.
6.
Carse, B., and Fogarty, T.C. (1996). Tackling the 'curse of dimensionality' of radial basis function neural networks using a genetic algorithm. In H-M Voigt, W. Ebeling, I. Rechenberg and H-P Schwefel (Eds.), Proceedings of the Fourth Conference on Parallel Problem Solving from Nature, Springer-Verlag Lecture Notes on Computer Science, vol.1141. pp.(710-719).
7.
French, R.M. and Messinger, A. (1994). Genes, phenes and the Baldwin effect: Learning and evolution in a simulated population. In R.A. Brookes and P. Maes (Eds.), Artificial Life IV: Proceedings of the Fourth International Workshop on Synthesis and Simulation of Living Systems. pp.(277-282). Cambridge MA: MIT Press.
8.
Gruau, F. and Whitley, D. (1993). The cellular development of networks: The interaction of learning and evolution. Research Report No. 93-04, Lyon: Ecole Normale Supérieure, Laboratoire de l'Informatique du Parallélisme.
9.
Harvey, I. (1996). Relearning and evolution in neural networks, Adaptive Behaviour, 4(1), 81-84.
10.
Harvey, I. and Stone, J.V. (1996). Unicycling helps your French: Spontaneous recovery of associations by learning unrelated tasks. Neural Computation , 8, 697-704.
11.
Harvey, I. (1997). Is there another new factor in evolution? Evolutionary Computation, 4(3), 313-329.
12.
Hinton, G.E. and Nowlan, S.J. (1987). How learning can guide evolution. Complex Systems, 1, 495-502.
13.
Hinton, G.E. and Sejnowski, T. (1986). Learning and relearning in Boltzmann machines . In D. Rumelhart, J. McClelland and the PDP Research Group (Eds.), Parallel Distributed Processing: Explorations in the Microstructure of Cognition. Volume 1: Foundations, pp.(282-317). Cambridge, MA: MIT Press/ Bradford Books.
14.
Lloyd Morgan, C. (1896). On modification and variation. ScienceIV (99), November 20, 1896. pp.(733-740).
15.
Mayley, G. (1996). The Evolutionary cost of learning. In P Maes, M. Mataric, J. Meyer, J. Pollack and S. Wilson (Eds.). From Animals to Animats 4: Proceedings of the 4th International Conference on Simulation of Adaptive Behaviour. pp.(458-467). MIT Press.
16.
Mayley, G. (1997). Landscapes, learning costs and genetic assimilation . Evolution, Learning and Instinct: A Special Edition of Evolutionary Computation4(3), 213-234.
17.
Nolfi, S., Elman, J. and Parisi, D. (1994). Learning and evolution in neural networks, Adaptive Behavior, 3, 5-28.
18.
Nolfi, S. (1999). How learning and evolution interact : The case of a learning task which differs from the evolutionary task. Adaptive Behaviour, 7(2), 231-236.
19.
Osborn, H.F. (1896). A Mode of evolution requiring neither natural selection nor the inheritance of acquired characters. Transactions of the New York Academy of Sciences, 15, 141-142, 148.
20.
Parisi, D., Nolfi, S. and Cecconi, F. (1992). Learning, behavior and evolution. In F.J. Varela & P Bourgine (Eds.), Toward a Practice of Autonomous Systems: Proceedings of the First European Conference on Artificial Life. Cambridge MA: MIT Press/Bradford Books.
21.
Parisi, D. and Nolfi, S. (1996). The Influence of learning on evolution. In R.K. Belew & M. Mitchell (Eds.), Adaptive Individuals in Evolving Populations, SFI Studies in the Sciences of Complexity, Vol. XXVI. pp.(419-428). Addison-Wesley
22.
Waddington, C.H. (1942). Canalization of development and the inheritance of acquired characters. Nature3811, November 14, 1942, pp.(563-565).
23.
Whitley, D., Scott Gordon, V. and Mathias, K. (1994). Lamarckian evolution, the Baldwin effect and function optimization. In Y. Davidor, H-P. Schwefel and R. Männer (Eds.), Proceedings of the Third Conference on Parallel Problem Solving from Nature, Springer-Verlag Lecture Notes on Computer Science, vol.866, pp.(6-15). Springer Verlag..
