1. Computational Worldview and the Sciences Leslie Valiant Harvard University

2. Potential Potential Disruptivity Supportivity Index PDI+PSI=10 Index (PDI) (PSI) 2? Computational Game Theory 8? Social Networks Computational Genomics Systems Biology 8? Quantum Computation 2? Neuroscience Evolution

3. THESIS Some questions in Biology are best formulated and analyzed in terms of appropriate Computational Models. (c.f. Turing, “Intelligent Machinery”, 1948)

4. 1. EVOLUTION How can biological evolution, in only a few billion years, give rise to such complex mechanisms?

5. 2. THE BRAIN How can brains, given such severe internal communication constraints, perform a sufficient set of primitives for general information processing?

6. Evolution Supposition that the eye could evolve “… absurd in the highest possible degree … …. eye does vary ever so slightly …” C. Darwin, Origin of Species, 1859. ??? Search for 3*10 9 length string in 3*10 9 years????

7. Previous Approaches to Evolution 1.Population dynamics of genes. 2. Evolutionary computation – algorithms for optimization or learning - inspired by biology, but not necessarily constrained by it.

8. Evolvability Mechanisms complex if they need different behaviors on different combinations of conditions. Represent as multi-argument functions. How can mechanisms for multi-argument functions evolve from ever so slightly simpler ones in a biologically plausible manner?

9. Inputs inputs output

10. Inputs inputs output

11. Inputs inputs output

12. Inputs inputs output

13. Learning e.g. Perceptron Hypothesis: 3x + 4y -7z > 0. Example: ( 1, 2, 0) is negative. Updated hypothesis: 2x +2y -7z > 0.

14. Evolution of Complex Mechanisms Evolution regarded as a constrained form of learning. f(x 1, …, x n ) is a target ideal function. It takes values on all combinations of inputs that are most beneficial. r(x 1, …, x n ) is a representation of a tentative hypothesis for f. Can r evolve towards f inexorably? Assume x i, f, r  {-1, 1}, x from distr. D.

15. Mutation is beneficial if performance increases by at least tolerance t(1/n,  ). It is neutral if tolerance does not go down by more than t, and not beneficial. If beneficial steps available one taken with relative probability, else if neutral steps available one taken with rel. prob. (Some robustness in defn.)

16. Boolean Function Classes Variables x 1, …, x n : 1.Disjunctions: e.g. x 1  x 3  x 7. True if at least one variable true. 2.Parities: e.g. x 1  x 3  x 7. True if an odd number are true.

17. Results Theorem 1: Parity functions are not evolvable over uniform distribution. Proof: 1.Evolvability  SQ learnability. 2. Parity  SQ [Kearns 98]. Theorem 2: Monotone conjunctions and disjunctions are evolvable over the uniform distribution.

18. Interpretation Theory describes granularity of evolvability towards targets relative to current modules. As modules for more targets evolve, further targets potentially become evolvable relative to these. Evolvability a constrained form of learnability. Evolution not more mysterious than learning.