What Does Artificial Life Tell Us About Death?

What Do es Artificial Life T ell Us Ab out Death? Carl o s Gershenson 1 , 2 , 3 1 Institut o de In v estig aciones en Matem´ atica s Apli cadas y en Sistema s Universidad Naciona l Aut ´ onoma de M ´ exico Ciudad Universitaria Ap do. Postal 20-72 6 / A dm n No. 20 0100 0 M´ exico D. F. M´ exi co cgg@unam .mx http ://tu ring .iimas.unam.mx/ ~ cgg 2 Cen tr o de Ciencias de la C ompleji dad Universidad Naciona l Aut ´ onoma de M ´ exico 3 Cen tr um Leo Ap ost el , V rije Universiteit Brussel Krijgsk u nd est raat 33 B-116 0 Brussel, Belgi um cgershen@vub. ac.b e ht tp:// home pages.vub.ac.be/ ~ cger shen September 21, 201 8 Every evil le aves a sorr ow in the memory, until the supr eme evil, de ath, wip es out al l memories to gether with al l life. –L e onar do da Vinci One of the op en problems in artificial life discuss ed b y Bedau, et al. ( Bedau et al. , 2000 ) is the establishmen t of ethical principles for artificial life. In particular: Muc h of curren t ethics is based on the sanctit y of h uman life. Researc h in articial life will affect our understanding of life and death (...) This, like the theory of ev olution, will hav e ma jor so cial consequences for human cultural practices suc h as religion. ( Bedau et al. , 2000 , p. 375) F o cussing o n our unde rstanding of death, this will dep end necessarily on our under- standing of life, and vice vers a. Throughout history t here ha ve b een sev eral explanations to b oth life and death, and it seems unfeasible that a consensus will b e reac hed. Th us, w e a re faced with m ultiple not ions of life, whic h imply differen t notions of death. Ho wev er, gener- ally sp eaking, if w e describ e lif e as a pro cess, death can b e understo o d as the irrev ersible termination of that pro cess. The general notion of life as a process or organizatio n ( Langton , 1989 ; Stereln y and Griffiths , 1999 ; Korzeniewsk i , 2001 ) has expelled vitalism from scien tific w orldviews. Moreo ve r, there 1 are adv antages in describing living systems from a functional p ersp ectiv e, e.g. it mak es the notion of life indep endent of its implemen t a tion. This is crucial for ar tificial lif e. Also, w e kno w that there is a constan t flow of matter and energy in living systems , i.e. their phy sical comp onen ts can c hange while the iden tit y of the organism is preserv ed. In this resp ect, one can mak e a v aria tion of Kauffman’s “blender thought exp erimen t” ( Kauffman , 2000 ): if y ou put a macroscopic living system in a blender and press “on”, a fter some seconds y ou will ha v e the same molecules that the living system had. How ev er, the org a nization of the living system is destroy ed in the blending. Th us, life is an orga nizatio nal asp ect of living systems, not so m uc h a physic al asp ect. Death o ccurs when this orga nization is lo st. One of the main properties of living organization is its self-pro duction ( V arela et al. , 1974 ; Maturana and V ar ela , 1980 , 198 7 ; Luisi , 1998 ; Kauffman , 20 0 0 ). When death o ccurs, this self-pro duction cannot b e maintained . But is this organizatio n the o nly thing that is lost with death? The no t ions of lif e and death hav e b een muc h related to those of mind, cognition, a w are- ness, consciousness, and soul 1 . On the one hand, the mind is a prop erty closely related with life. Some prop o se that mind and life are essen tially the same pro cess ( Stewart , 199 6 ; Bedau , 1998 ). On the o ther hand, p eople hav e sp eculated since the da wn of civilization on what o ccurs with the mind after death. Life is a pro cess described b y a n observ er ( Maturana and V arela , 1 987 ), in first or third p erson p ersp ectiv e. When the pro cess breaks, only description in the third p erson observ er remains. By definition, w e can only sp eak ab o ut death fro m a third p erson p ersp ectiv e. What can artificial lif e add to this discussion? Artificial life simulations (“soft” ALife) can b e seen as opa que t ho ugh t exp erimen ts ( Di P aolo et al. , 200 0 ), i.e. one can explore differen t notions of life and death with them. Rob ots (“ ha rd” ALife) w ould also serv e this purp ose. Artificial life can help us build living systems to b e explored from a third p erson p ersp ectiv e in a syn thetic w ay ( Steels , 1993 ). Can w e say that “animats” ( Wilson , 1985 ) ha v e a mind, in the same sense as animals do? If not, is there something missing in the particular animat, in artificial life, or in the observ er? When a digita l organism dies ( Ra y , 1994 ), what ph ysically c hanges is the RAM that enco ded the organism in bits. When the bits describing the organization of the org a nism are erased, t he only pla ce where t he organism prev ails is in the observ er. The same is for rob o t s. The same is for animals. The same is for h umans. If w e describ e life as an organizationa l pro cess, and a mind as dep ending on it ( Clark , 1997 ), when the organization is lost, the life is lost and the mind is lost. Certainly , the organization of digit al organisms is m uc h easier to preserv e than that of biological ones. Apa r t from the ease o f copying digital informatio n, digital organisms are generally inhabiting closed environme nts. Biological org anisms face op en dynamical en viron- men ts that constan tly threat en their integrit y . i.e. organisms need to mak e thermo dynamical w ork (metab olism) ( Kauffman , 20 00 ) to main tain themselv es. In an op en en vironmen t suc h as the biosphere, where different ev olving org a nisms interact, there is no “b est” or “fitt est” organism, since the fitness dep ends on the dynamic env ironment. Th us, fitness changes con- stan tly with the environme nt, since the en vironmen t is c hanged as organisms ev olv e trying to increase their fitness. In this contex t, it can b e sp eculated that there is an ev olutionary 1 It is quite problema tic to attempt to define these, but a v ague notion will suffice. In the following, “mind” will b e us e d in a bro ad s e nse that includes a lso cognition, aw arene s s, conscio usness, and soul. 2 adv antage of death. If there was no death, i.e. if an o r ganism somehow managed to main- tain its organization indefinitely , ev olution w ould stop. This a ctually o ccurs commonly with digital ev olution. In fact, death of digital or g anisms has b een used as a measure to intro duce no v elt y ( Ra y , 1994 ; Dorin , 2005 ; Olsen et al. , 200 8 ). The organization represen ted in a sp ecies can surviv e sev eral lifespans, but is sub ject to the same pressure as the one just describ ed. Th e loss of organization giv es the o pp ortunit y to new forms of organization to dev elop. The dev elopmen t of proto cells ( Rasm ussen et al. , 20 0 8 ) (“w et” ALife) might further con- tribute to the exploration of the notio ns of life a nd death. By che mically pro ducing the organization resulting in living systems , the non-m ystical notio n of life review ed here will gain further gr o unds. Additionally , the non-m ystical notion of death explored here will hav e to b e further elab orated. What o ccurs when a proto cell dies? If w e can create again a living system with the same organization, did it die in the first pla ce? W e will b e able to ha v e differen t instantiations of the same living o rganization, j ust like w e can hav e differen t copies of the same digital organism. Will its death ha v e the same meaning as that of an animal? One thing to notice in these questions is that in mo st biological organisms, the organi- zation lies not only in their genes, but also in their deve lopment (epigenesis). Clones can dev elop different organizations. The same might o ccur for proto cells and other future “ w et” artificial living systems. Ho w ev er, on the digital side of artificial life, it is easy again t o main tain and repro duce the organization acquired t hrough dev elopmen t ( Balk enius et al. , 2001 ). What will the future bring? Will there b e biolog ical systems closer to digita l ones, in the sense that living info rmation can b e maintained and/or repro duced? Probably . Ho w will this affect death? W e will hav e more con trol o v er it . Will this mark an end to ev olution? No, ev en when some living organization migh t b e more p ersistan t, there will a lw a ys b e new situations where organisms ha v e to adapt. In an y case, the cultural attitudes tow a rds death most probably will change. This is not suggesting t hat w e will b e less touch ed b y it , o r less spiritual tow ards it. The implication is tha t w e will ha v e a b etter understanding of the phenomenon, with a broader scien tific ba sis. T o conclude this philosphical essa y , different notions of death will b e deduced fro m a limited set of differen t notio ns of life: • If we consider life as self-pro duction ( V arela et al. , 1974 ; Maturana and V arela , 1980 , 1987 ; Luisi , 1998 ), then death will the the loss of that self-pro duction a bility . • If w e consider life as what is common to all living b eings ( De Duv e , 2003 , p. 8), t hen death implies the termination of that commonalit y , distinguishing it from o ther living b eings. • If w e consider life a s computation ( Hopfield , 1994 ), then death will b e the end (halting ?) of that computing pro cess. • If w e consider life as supple adaptation ( Bedau , 1 9 98 ), death implies the lo ss of that adaptation. 3 • If w e consider life as a self-repro ducing system capable o f at least o ne thermo dynamic w ork cycle ( Kauffman , 2000 , p. 4) , death will o ccur when the system will b e unable to p erform thermo dynamic work. • If w e consider life as info r ma t io n (a system) that pro duces more of its ow n info rmation than that pro duced by its environme nt ( Gershenson , 20 0 7 ), then death will o ccur when the en vironmen t will pro duce more information tha n tha t pro duced b y the system. References Balkenius, C. , Zla tev, J. , Brezeal, C. , Da utenhahn, K. , and Kozim a, H. , Eds. (2001). Pr o c e e dings of the First International Workshop on Epigenetic R ob otics: Mo deling Co gnitive Developme nt in R ob otic Systems . V o l. 85. Lund Univ ersit y Cognitiv e Studies, Lund, Sw eden. URL http://www. lucs.lu.se/LUCS/085/ . Beda u, M. , McCaskill, J. , P ackard, P. , Rasmusse n, S. , Gre en, D. , Ikegami, T. , Kaneko , K. , and Ra y, T. (20 00). Op en Problems in Artificial Life. Artificial Life 6 ( 4): 363–376. Beda u, M. A. (1998). F our puzzles a b out life. Artificial Life 4 : 125–140. URL http://tiny url.com/3a59sy . Clark, A. (1997 ). B eing Ther e: putting br ain , b o dy, and world to gether again . MIT Press, Cam bridge, MA. De Duve, C. (2003). Live Evolving: Mole cules, Mind, and Me aning . Oxford Univ ersit y Press. Di P aolo, E. A. , Noble, J. , and Bullock, S. (2000). Sim ulation models as opaque though t exp erimen ts. In Art ificia l Life VII: Pr o c e e dings of the Seventh In- ternational Confer e nc e on A rtificial Lif e , M. A. Bedau, J. S. McCaskill, N. H. P ac k ard, and S. Rasm ussen, (Eds.). M IT Press, Cam bridge, MA, pp. 497–5 0 6. URL http://epri nts.ecs.soton.ac.uk/11455/ . Dorin, A. (2 005). Artificial life, death and epidemics in ev olutionary , generativ e electronic art. In Applic a tion s on Evolutionary Computing . Lecture Notes in Computer Science, vol. 3449/200 5. Springer, Berlin, Heidelb erg, pp. 448–457 . URL http://is.g d/VOOO . Gershenson, C. (2007). The world a s ev olving information. In Pr o c e e dings of In- ternational Conf er enc e on Com plex Systems IC CS2007 , Y. Bar-Y am, (Ed.). URL http://uk.a rxiv.org/abs/0704.0304 . Hopfield, J. J. (19 9 4). Ph ysics, computation, and wh y biology lo oks so differen t. Journal of The or etic al Biolo g y 171 : 53–6 0 . URL http://dx.d oi.org/10.1006/jtbi.1994.1211 . Kau ffman, S. A. (2000) . Investigations . Oxford Univ ersit y Press. 4 K orzeniewski, B. (2001) . Cyb ernetic fo r m ulation of the definition of life. Journal of T h e o- r etic al Biol o gy 209 (3): 27 5 – 286. URL http://dx.doi.org /10.1006/jtbi.2001.2262 . Langton, C. (1989). Artificial life. In A rtificial life , C. Langton, (Ed.). San ta F e Institute Studies in the Sciences of Complexit y . Addison-W esley , Redwoo d Cit y , CA, pp. 1–47. Luisi, P. L. (1998). Ab out v arious definitions of lif e. Origins of Life and Evolution of Biospher es 28 ( 4 -6) (Octob er): 613–622. Ma turana, H. and V arela, F. (1980). Autop oiesis and C o gnition: the R e alization of the Living , 2nd ed. D. Reidel Publishing Co., Dor dech t. Ma turana, H. R. and V arela, F. J. (198 7). T he T r e e of Know le dge: The Biolo gic a l R o ots of Human Understand ing . Sham bhala. Olsen, M. , Siegelmann-Danieli, N. , and Sie gelmann, H. (200 8). Robust ar t ificial life via artificial prog r a mmed death. Artificial In tel ligenc e 172 (6-7 ): 8 8 4 – 89 8. URL http://dx.d oi.org/10.1016/j.artint.2007.10.015 . Rasmussen, S. , Beda u, M. A. , Chen, L. , Deamer, D. , Krakauer, D. C. , P ackard, N. H. , and St adler, P. F. , Eds. (20 0 8). Pr oto c el ls: Bridging Nonlivin g and Living Matter Bri d g ing Nonlivi n g and Living Matter . MIT Press. Ra y, T. S. (1994). An ev olutionary approach to syn thetic biology: Z en and the art of creat- ing life. A rtificial Life 1 (1/2): 1 9 5–226. URL http://www. his.atr.jp/ ~ ray/pubs/ze n/ . Steels, L. (199 3 ). Building agen ts out of autono mo us b eha vior systems. In Th e Artificial Life R oute to A rtificial In tel ligenc e: Building Emb o die d Situate d A gents , L. Steels and R. A. Bro oks, (Eds.). La wrence Erlbaum. Sterelny, K. and Griffiths, P. E. (1999 ). Sex and de a th . Univ ersit y of Chicago Press. Stew ar t, J. (199 6 ). Cognition = life: Implications for higher-lev el cognitio n. B e haviour al Pr o c esses 35 (1 – 3): 311 –326. V arela, F. J. , Ma turana, H. R. , and Uribe., R. (1974). Autop oiesis: The orga nization of living systems , its characteriz atio n and a mo del. BioSystems 5 : 187– 196. Wilson, S. W. (1985 ) . Knowledge gr o wth in an art ificial animal. In Pr o c e e dings of the First International Confer enc e on Genetic Algorithms and Their Applic ations . Lawrence Erlbaum Asso c, Hillsdale, NJ, pp. 16 – 23. 5