Place In Mind: Towards a Dynamic Memory Palace




This paper will dis­cuss in­ter­dis­ci­pli­nary work on mem­ory be­tween psy­chol­o­gists, bi­ol­o­gists, ro­boti­cists, neu­rol­o­gists, lin­guists and dis­cuss ar­che­typ­i­cal mod­els gleaned from those dis­ci­plines. This will in­clude his­tor­i­cal mem­ory palaces, neural net­works, lin­guis­tic struc­tures, and bi­o­log­i­cal sys­tems.


Mnemonic de­vices: Men­tal Maps and Mem­ory Lanes
Using the In­ter­net to focus on pin­point­ing par­tic­u­lar nuggets of knowl­edge, while sub­merged in an il­lu­sion of an in­fi­nite amount of data ac­ces­si­ble through ran­dom ac­cess, puts for­ward the idea of in­fi­nite com­bi­na­tions of data and ac­cess paths. This com­bi­na­tory power, along with how a user’s choices are in­formed and fil­tered, erodes the edges of pre-ex­ist­ing con­sen­sus mod­els, laws and iden­ti­ties. Bor­ders be­tween dis­ci­plines and par­a­digms are ex­pe­ri­enced on the web as in­creas­ingly ten­u­ous, ar­bi­trary and dy­namic.  What kind of model, if any, can serve to sup­ply some con­straints of struc­ture on a net­worked art work.

The Ad Heren­nium (circa 86-82 B.C.), a text­book on rhetoric, con­tained a mem­ory sec­tion di­vided into “rules for places, rules for im­ages, mem­ory for things, mem­ory for words.” In­struc­tions were to first fix places based on the con­struc­tion of men­tal ar­chi­tec­tural mod­els. The larger and more com­plex the bet­ter be­cause more mem­o­ries could be housed. Knowl­edge was bro­ken down into dis­crete par­ti­cles and topog­ra­phy con­structed as a mem­ory strat­egy. Bits of in­for­ma­tion were as­signed to ob­jects; ob­jects were placed in spe­cific rooms, in a spe­cific path, which would lead the re­caller to data in the cor­rect order. It was im­per­a­tive that knowl­edge was fixed – ad­hered to an ob­ject — so that it could be ac­cessed at any time by vir­tu­ally beat­ing a path to its door.

Both stor­age and mem­ory both par­tic­u­larly played a large part in the con­cerns of me­dieval schol­ars, the­olo­gians, sci­en­tists and artists. In me­dieval West­ern Eu­rope the ar­ti­fi­cial mem­ory struc­ture par­a­digm shifted from ar­chi­tec­tural mod­els to the­o­log­i­cal con­structs. The ter­ri­to­ries of heaven and hell be­came host to spe­cific mem­o­ries. Mem­o­ries were as­signed to ob­jects, crea­tures and topo­graph­i­cal land­marks in those ter­ri­to­ries. The­o­log­i­cally based mem­ory palaces had a large in­flu­ence on cre­ative forces of their times. It is pos­si­ble to view Dante’s In­ferno as a jour­ney through what would have been a per­va­sive and pop­u­lar mem­ory palace par­a­digm. In The Art of Mem­ory, Fran­cis A. Yates de­scribes the In­ferno as “based on or­ders of places Hell, Pur­ga­tory and Par­adise…. The Di­vine Com­edy would thus be­come a summa of simil­i­tudes and ex­am­ples, with mem­ory as the con­vert­ing power, the bridge be­tween the ab­strac­tion and the image.”

In the 16th cen­tury, the mem­ory palace emerged from its vir­tual state into a phys­i­cal one ex­ist­ing out­side of the mind in the form of an in­stal­la­tion. The Mem­ory The­atre of Giulio Camillo In­ter­po­lated the Greek mem­ory palace by con­struct­ing a wooden struc­ture that be­came the in­spi­ra­tion for the ar­chi­tec­ture of Shake­speare’s Globe the­ater . Camillo’s the­ater was a rep­re­sen­ta­tion of the uni­verse ex­pand­ing its in­cep­tion through the stages of cre­ation. Every per­son who en­tered this mag­i­cal por­tal would come away ca­pa­ble of speak­ing on any sub­ject with the skill of Ci­cero. The wooden struc­ture, large enough for two peo­ple, was com­mis­sioned by the King of France and dis­played in Venice and Paris.

A re­con­struc­tion of The Mem­ory The­ater by Frances Yates

“The work is of wood, marked with many im­ages, and full of lit­tle boxes; there are var­i­ous or­ders and grades in it. . . . He calls this the­atre of his by many names, say­ing now that it is a built or con­structed mind and soul, and now that it is a win­dowed one. He pre­tends that all things that the human mind can con­ceive and which we can­not see with the cor­po­real eye, after being col­lected to­gether by dili­gent med­i­ta­tion may be ex­pressed by cer­tain cor­po­real signs in such a way that the be­holder may at once per­ceive with his eyes every­thing that is oth­er­wise hid­den in the depths of the human mind.”


A 20th cen­tury labyrinthine line draw­ing by Um­berto Eco (http://​www.​intelligentagent.​com/​archive/​RoadEco.​gif) traces the de­vel­op­ment of a pun cre­ated by James Joyce for Finnegan’s Wake. The draw­ing il­lus­trates a de­cod­ing of the pun, trac­ing pos­si­ble nodes of as­so­ci­a­tion which link the words “Ne­an­derthal,” “Me­an­der,” and “Tale” from which Joyce formed the trans­for­ma­tive word “Me­an­der­tale.” The newly con­structed word “Me­an­der­tale” ap­pears to sig­nify the very name of the process that forms it, a me­an­der­ing quest for as­so­ci­a­tions be­tween words–a quest where these as­so­ci­a­tions si­mul­ta­ne­ously tell the story of the words’ evo­lu­tion and trans­form them. In such a well-ven­ti­lated world, per­haps one nec­es­sary con­straint might be to as­sume that no word suf­fers more than six de­grees of sep­a­ra­tion from any other.

Clues With­out Con­text

How far astray can the routers of imag­i­na­tion take us from what we can con­sider true? Can we be sure that such a quest on a well-trod­den path is re­veal­ing good metaphor­i­cal ex­pres­sion? Clues with­out con­text i.e., the nodes with­out synapses–make the sub­ject of the di­a­gram cease to res­onate its po­ten­tial; it sim­ply be­comes a rep­re­sen­ta­tion of it­self where every­thing is equal or in a sim­plis­tic hi­er­ar­chy (some are upper case, some are lower). The labyrinthine jour­ney de­fined by the con­vo­luted node line oc­cur­ring be­tween words is sprin­kled with al­chem­i­cal events: oc­cur­rences in lan­guage where two words to­gether form an­other that leads to a whole new ex­pres­sion of as­so­ci­a­tion and mean­ing. Such events stim­u­late ver­nac­u­lar, slang and new lan­guages.

Mys­tery With­out [E]mo­tion

Then again, se­lect­ing the nodes in Eco’s draw­ing with­out their names or con­nec­tions of­fers mys­tery (go ahead, con­nect the dots)–but the lit­er­ary metaphor suf­fers a loss of its mus­cu­lar tone, its life-like ex­u­ber­ance. As the de­pic­tion of con­vo­luted con­nec­tions dis­ap­pear in the above draw­ing, the res­o­nance also falls away, dwin­dling to noth­ing. If we ab­sent every­thing but the line, its func­tion as a pointer to pos­si­bil­i­ties suf­fers. We have uni­formly sub­tracted from this path any of its sig­nif­i­cant el­e­ments.
Order with­out sig­nif­i­cance

Net­worked art pre­sents a process than a prod­uct. But the process alone with­out the nar­ra­tive, or the sig­nif­i­cant rea­son to travel through it, seems bar­ren.   Where the re­la­tion­ship of one bit of con­tent to an­other in these works, even at a min­i­mum con­jures up ad­di­tional lan­guage in the vis­i­tor’s mind that makes the leap from one work to an­other. This in­vis­i­ble text, an emerg­ing hi­ero­glyphic struc­ture in the reader’s mind, is the ac­tiv­ity that con­jures up a new lan­guage of the links and a new mode of pre­sen­ta­tion. Hy­per­me­dia links are pock­ets, ab­sences, lapses, and synapses, in­di­cat­ing what is in­ex­press­ible or in­ter­ac­tively as­sumed by the viewer or reader. The sub­text of the work can be em­bed­ded, even in­con­sis­tently, in the link­ages and left to the viewer to de­ci­pher.

“Plan­ning is just a way of avoid­ing fig­ur­ing out what to do next”

The prob­lems of spa­tially aware em­bod­ied cog­ni­tion are per­va­sive. Many dis­ci­plines have had to de­velop strate­gies to deal with basic ques­tions of ‘Where am I?’, ‘Where is here?’ and ‘How do I nav­i­gate through this place?’ in many forms. Some par­tic­u­larly promis­ing tech­niques have been ex­plored at the in­ter­sec­tions be­tween the fields of cog­ni­tive sci­ence and ro­bot­ics that may be use­ful for de­vel­op­ing the kinds of ad­vanced spa­tial nav­i­ga­tion of data spaces pro­posed in this paper.  Ro­bots have be­come ad­vanced enough and cheap enough to be em­ployed in a wide va­ri­ety of gen­eral pur­poses in rel­a­tively un­con­trolled en­vi­ron­ments ( for ex­am­ple: using an iRo­bot Roomba(tm) robot vac­uum cleaner to clean the liv­ing room floor while you are at work with­out hav­ing to fear for the safety of any house­hold pets that may be roam­ing about) in­stead of being re­stricted to use in con­trolled in­dus­trial en­vi­ron­ments such as au­to­mated fac­to­ries.

These sys­tems adap­tively and ef­fi­ciently  ex­plore their en­vi­ron­ment using lim­ited con­tact sen­sors to de­ter­mine when they col­lide with walls or other ob­jects as they fol­low a semi ran­dom­ized path through the space. They lack ad­vanced com­puter vi­sion tech­niques and have no pre-ex­ist­ing model of what the space is like. They sim­ply know which way they are head­ing and when they col­lide with some­thing that would pro­hibit them mov­ing for­ward, they go an­other di­rec­tion and change the method the pat­tern they were using to tra­verse the room (al­ter­nat­ing from mov­ing from side to side, spi­ral­ing about the room and doing ran­dom walks). This may sound sim­ple, be­cause it is.

The robot vac­uum doesn’t build a high level com­plex model of the space, it sim­ply re­acts to what is im­me­di­ately hap­pen­ing to it and tries dif­fer­ent things when what it was doing did not work. It does not do that be­cause it does not need to.  It is able to sense where the bound­aries of the area it wishes to ex­plore are and then en­gages in a con­trolled and pur­pose­ful wan­der­ing of that space. The di­rect feed­back that it gets from its en­vi­ron­ment sub­sti­tutes for a com­plex con­cep­tual model. It is a re­ac­tive sys­tem, with rel­a­tively lit­tle in­ter­nal vari­able state to model the en­vi­ron­ment. This ap­proach grew out of bi­o­log­i­cal mod­els  of how ants and other in­sects nav­i­gate over large areas by fol­low­ing ex­tremely sim­ple rules that rely on re­spond­ing di­rectly to the en­vi­ron­ment more than they do on build­ing any com­plex model of the en­vi­ron­ment . This sub­sump­tion ar­chi­tec­ture of in­te­grated sim­ple be­hav­iors that re­spond to a rich en­vi­ron­ment has been used in a va­ri­ety of areas of ro­bot­ics and for tasks such as sim­u­lat­ing ap­par­ently com­plex human like be­hav­ior in the Sims se­ries of video games . They pre­sent a pos­si­ble method for ex­plor­ing and de­sign­ing the sorts of rich data spaces that are pro­posed in this paper.

Ref­er­ences and Notes:

 The Art of Mem­ory, Yates, Frances A., Pim­lico Press, Lon­don, first pub­lished 1992, 2000 ed. Cited.,pp.20-21. Ibid.,pp.104-105.


Ibid, Chap­ter 6. Eras­mus, “Epis­to­lae, ed. P.S. Allen and oth­ers, IX, p. 479, let­ter from Vig­ilus to Eras­mus, as quoted by Frances Yates in the Art of Mem­ory, pp. 136-137.


Di­a­gram by Um­berto Eco from The Lim­its of In­ter­pre­ta­tion, A Mid­land Book, In­di­ana Uni­ver­sity Press, Bloom­ing­ton and In­di­anapo­lis, 1994, Il­lus­tra­tion 9.2, p. 141.



R. A. Brooks (1987). “Plan­ning is just a way of avoid­ing fig­ur­ing out what to do next”, Tech­ni­cal re­port, MIT Ar­ti­fi­cial In­tel­li­gence Lab­o­ra­tory.

Rod­ney Brooks (Jan­u­ary 1991), “In­tel­li­gence with­out rep­re­sen­ta­tion”, Ar­ti­fi­cial In­tel­li­gence 47 (1-3): 139–159, doi:10.1016/0004-3702(91)90053-M

Steels, Luc & Brooks, Rod­ney, ed. (1995), The Ar­ti­fi­cial Life Route to Ar­ti­fi­cial In­tel­li­gence: Build­ing Em­bod­ied, Sit­u­ated Agents, Hills­dale, New Jer­sey: Lawrence Erl­baum As­so­ci­ates, ISBN 0-8058-1519-8

For­bus, Ken (2001) “Under the hood of the sims” http://​www.​cs.​northwestern.​edu/​~for­bus/c95-gd/lec­tures/The_Sim­s_Un­der­_the_­Hood_­files/v3_­doc­u­ment.htm

Comments are closed.