A apresentação está carregando. Por favor, espere

A apresentação está carregando. Por favor, espere

Epistemologia e a Filosofia da Mente Incorporada e Situada

Apresentações semelhantes


Apresentação em tema: "Epistemologia e a Filosofia da Mente Incorporada e Situada"— Transcrição da apresentação:

1 Epistemologia e a Filosofia da Mente Incorporada e Situada
Pim Haselager Dpt. of Artificial Intelligence/Cognitive Science, NICI, University of Nijmegen, The Netherlands Post-Graduate Program in Cognitive science and Philosophy of Mind, Philosophy Department, UNESP, Marilia, SP, Brazil

2 Ciência cognitiva Psicologia Inteligência artificial (robótica)
Filosofia Neurociência Lingüística Biologia Etologia Antropologia

3 Comportamento e conhecimento comum
Comportamento racional (normalmente coerente) Conhecimento descritivo Sobre o mundo Atitudes proposicionais Raciocinar e planejar na base do conhecimento proposicional

4 Racionalidade Wordnet dicionário:
A qualidade de ser consistente com a lógica A capacidade de ter bom senso Webster’s dicionário: a qualidade ou estado de ser razoável: concordância com razão Coerência do conteúdo, não aleatório

5 Os seres humanas são racionais?
A filosofia tradicional: sim, leis de lógica são as leis do pensamento

6 Tradição filosófica: Raciocinar é aplicar a lógica
Hobbes: reasoning is but reckoning Leibniz: characteristica universalia. Boole 1854: Laws of logic: "An investigation of the laws of human thought on which are founded the mathematical theories of logic and probabilities". J.S. Mill 1843: A system of Logic Na ciência cognitiva: Braine 1978; Rips 1983; Braine & O’Brien, 1991 Mensen bezitten + elementaire abstracte regels (bv. modus ponens) v syntactische aard over prop.repr. -> niet domein-specifiek; ± 16 van dat soort regels, (exp. bewijs: single step tasks errorfree, samengestelde taken tijd & fouten analyse + strategieën die inferentieketens coördineren en regels selecteren.

7 Tabelas da verdade 1880’s: Peirce, Frege
P -> q is true when not p and q because we have no grounds for asserting that p->q is false. P->q material implication, p <->q: material equivalence or biconditional Schröder wordt ook wel genoemd ivm waarheidstafels.

8 Inferências validas Modus Ponens: Se p então q p Então q
Modus Tollens: Não q Então não p

9 Ciência Cognitiva Tradicional
O mundo Cognição Modelagem do mundo Conhecimento Crenças, opiniões e desejas Metas Raciocínio Planejamentos Decisões Sensores Efetuadores

10 Os termos primitivos dos paradigmas na robótica
Input Output Sentido Dados sensoriais Informação percebida Plano Informação percebida e/ou cognitiva Diretrizes (instruções) Functions that take information from sensors and produce output used by other functions Functions take information (from sensors or internal knowledge) produce tasks to perform (go down the hall, turn left for 3 meters and stop). Functions produce output commands to motor actuators (turn 90 degrees clockwise with 0.2mps) Ação Informação percebida e/ou diretrizes Comandos aos efetores

11 O paradigma hierárquico
Primitivos Input Output Sentido Dados sensoriais Informação percebida Plano Informação percebida e/ou cognitiva Diretrizes (instruções) Functions that take information from sensors and produce output used by other functions Functions take information (from sensors or internal knowledge) produce tasks to perform (go down the hall, turn left for 3 meters and stop). Functions produce output commands to motor actuators (turn 90 degrees clockwise with 0.2mps) Ação Informação percebida e/ou diretrizes Comandos aos efetores

12 O paradigma hierárquico
Sentido-Plano-Ação (Sense-plan-act) Uma modelagem do mundo A assumpção do mundo ‘fechado’ O modelo do mundo tem tudo que o robô precisa saber Raciocinar logicamente no base do conhecimento proposicional Agir só depois planejar , top down, heavy on planning, based on introspection, one global world model closed world assumption: the world model contains everything the robot needs to know, there can be no surprises frame problem: representing a real-world situation in a way that is computationally tractable (p.53).

13 Shakey 1960’s Controlada pelo Strips TV câmera
Percepção óptica da distancia Sensores de pára-choque 1970-Shakey the robot reasons about its blocks Built at Stanford Research Institute, Shakey was remote controlled by a large computer. It hosted a clever reasoning program fed very selective spatial data, derived from weak edge-based processing of camera and laser range measurements. On a very good day it could formulate and execute, over a period of hours, plans involving moving from place to place and pushing blocks to achieve a goal. Shakey the Robot Shakey was the first mobile robot to reason about its actions. Developed by SRI's Artificial Intelligence Center from 1966 through 1972 (we were called the Stanford Research Institute at that time), Shakey has had a substantial legacy and influence on present-day artificial intelligence and robotics. Shakey had a TV camera, a triangulating range finder, and bump sensors, and was connected to DEC PDP-10 and PDP-15 computers via radio and video links. Shakey used programs for perception, world-modeling, and acting. Low level action routines took care of simple moving, turning, and route planning. Intermediate level actions strung the low level ones together in ways that robustly accomplished more complex tasks. The highest level programs could make and execute plans to achieve goals given it by a user. The system also generalized and saved these plans for possible future use. SRI's mobile robot, Flakey, built in 1984 reflected the dramatic improvements 14 years makes in the state-of-the-art. It was approximately 3 feet high and 2 feet in diameter. Like Shakey, it had 2 drive wheels with encoders, but their maximum velocity was 2 feet per second instead of inches. The robot had 12 sonar  range finders for obstacle avoidance and navigation in real spaces, not just a laboratory like Shakey's. Again a video camera and laser provided range-finding information over a small area in front of Flakey. And Flakey's computers included one of the then new personal workstations, giving it far more intelligence than its predecessor, in far less space. Flakey was the first robot server to use Saphira, the high-level robotics development environment that has grown over decades into the powerful system employed by ActivMedia robots today. With Saphira, Flakey could navigate and recognize corridors, find doorways and plan a path from one place to another. Both Shakey and Flakey were large, one-of-a-kind robots costing many tens or hundreds of thousands of dollars to build. Shakey, a mobile robot developed at the Stanford Research Institute by Bertram Raphael, Nils Nilsson and their group, during the period (Shakey can now be viewed at the Boston Computing Museum.) The robot occupied a specially built micro-world consisting of walls, doorways, and a few simply-shaped wooden blocks. Each wall had a carefully painted baseboard to enable the robot to "see" where the wall met the floor (a simplification of reality that is typical of the micro-world approach). Shakey had about a dozen basic abilities, such as TURN, PUSH and CLIMB-RAMP. These could be combined in various ways by the robot's planning programs. Shakey's primary sensor was a black-and-white television camera. Other sensors included a "bump bar", and odometry that enabled the robot to calculate its position by "dead reckoning". A demonstration video showed Shakey obeying an instruction to move a certain block from one room to another by locating a ramp, pushing the ramp to the platform on which the block happened to be located, trundling up the ramp, toppling the block onto the floor, descending the ramp, and manoeuvring the block to the required room, this sequence of actions having been devised entirely by the robot's planning program without human intervention. Critics emphasise the highly simplified nature of Shakey's environment and point out that, despite these simplifications, Shakey operated excruciatingly slowly--the sequence of actions in the demonstration video in fact took days to complete. The reasons for Shakey's inability to operate on the same time-scale as a human being are examined later in this article.

14 Shakey, salas e caixas Shakey Sala3 Sala2 Sala1 Caixa2 Porta2 Porta1

15 O problema do ‘frame’ McCarthy & Hayes (1969): “queremos criar uma maquina que possa interagir de maneira inteligente com o mundo” O problema da representação do conhecimento comum Como manter a consistência do modelo do mundo O problema do uso eficiente do conhecimento relevante

16 Os seres humanas são racionais?
A filosofia tradicional: sim, leis de lógica são as leis do pensamento A psicologia e filosofia da mente incorporada e situada: Muitas vezes não, racionalidade limitada

17 A tarefa de Wason: a seleção de cartas
“Se existe uma vogal de um lado então haverá um numero par do outro lado” p: vogal, q: numero par, se p então q E K 4 7 Como detetar que o reglo e verdade? Procura falsificacao. K= consoane (consonant) Ivm 4: als een klinker prima, als een medeklinker, dan zegt de regel daar niets over. Klinker -> even getal. Welke moet je kiezen om de regel te testen: E (=p) want als niet even dan regel onjuist 7 (=not q) ook, want als klinker dan klopt regel niet K (=not p) niet, want even of oneven maakt niet uit, denial of the antecedent 4 (=q) niet, want e of k maakt niet uit, affirmation of the consequent 4: afirmação do conseqüente k: negação do antecedente E: se não há um numero par, a regra não é apropriada 7: se não há uma vogal, a regra não é apropriada

18 Wason’s tarefa: muito difícil
Johnson-Laird & Wason (1970) Só 5 de 128 pessoas escolheram correto (E e 7) 59: E e 4 42: E Contexto mais prático (cartas e selos) 92% coreto Contexto deontico (deve fazer isso) é mais fácil que o contexto indicativo (modus ponens) E K 4 7 “Se existe uma vogal de um lado então haverá um numero par do outro lado” Se 7 tiveria um vogal, o regre é erado (p=vogal, numero par=q). Escolhar 4 ( p->q, q) Afirmacao do consequente Talvez pessoas procuram confirmacao em vez de falsificacao. Mas no outras tarefas nao parece com isso (Eysenck & Keane, 2000, p.454). Deontic context: not p-> q, but if you do p then you must do q. Abstract is indicative, pratice is deontic. The memory cueing hypothesis: it is the specific experience with a realistic context that explains the context effects. However, now it seems that tasks that use a theme as a context are actually requiring dealing with regulations (do’s and don’ts): deontic reasoning. When the task is given in some deontic form, then facilitation is found, even for abstract versions of the task. Cheng & Holyoak (1985) gave subjects in Hong-kong ad in Michigan the same two versions of the task: the postal problem and checking the passengers forms at an airport (cholera). In H-K subjects are familiar with the postal rule, so they would be better there than subjects from M, but equivalent at the airport version. When given a rationale, that effectively changed the context to the deontic one, improvement in all versions both for H-K and M were expected. H-K was better than M at the postal (=memory-cueing), but faciliation occurred whenever a deontic context was provided.

19 Inferências invalidas
Afirmação do conseqüente Se p então q q Então p Negação do antecedente Não p Então não q If it's raining then the streets are wet. The streets are wet. Therefore, it's raining. Counter-Example to confirming the consequent: If it's snowing then the streets will be covered with snow. The streets are covered with snow. Therefore, it's snowing. Counter-example to deneying the antecendent: If it's raining, then the streets are wet. It isn't raining. Therefore, the streets aren't wet.

20 Seres humanas: lógico ou não lógico?
Pessoas acham muitas vezes que inferências invalidas são validas (Eysenck & Keane, 2000, p ) Marcus & Rips, (1979) 33% afirmação do conseqüente 21% negação do antecedente pode ser ate 70%, dependendo dos contextos

21 Perguntas Psicologia: Epistemologia e Filosofia da Mente
Como explicar os observações empíricas de (falhas no) raciocínio dos seres humanas? Epistemologia e Filosofia da Mente (Por)Que é: nossa cognição? para pensar, planejar e saber para (inter)agir com o mundo

22 Senso comum ou senso artificial?
“Examinemos um problema famoso do senso comum: que conhecimento está envolvido na formulação de um planejamento simples para quebrar um ovo em uma tigela, com a intenção de que o conteúdo do ovo esteja dentro da tigela?” (Elio, 2002, p.11). (Elio, R. (2002). Common sense, reasoning & rationality. Oxford: Oxford University Press, p.11). “Consider a famous commonsense problem: what knowledge comes into play in formulating a simple plan to crack an egg on the side of a bowl, with the intention that the egg contents will be in the bowl?” I think that the formulation of this ‘common sense’ problem is everything but common sense. That is, assuming that we use (sub-consciously) explicitly represented knowledge in order to formulate an explicit plan may be on the wrong track altogether. Indeed, one of the more important problems that cognitive science is struggling with is how to make computers that display common sense and that can perform everyday tasks.

23 A falta de entendimento do comportamento comum
A formulação do problema está errada Suposições erradas: O conhecimento comum consiste de atitudes proposicionais O comportamento é uma conseqüência dos processos de inferências dedutivas com base no conhecimento proposicional sobre o mundo O comportamento é uma conseqüência dos planejamentos explícitos IA tradicional está olhando na direção errada

24 Embodied Embedded Cognition Cognição Incorporada e Situada
A interação corporal com o meio ambiente é de fundamental importância e não de importância secundária para a cognição Títulos: enactive cognition, situated cognition, embedded cognition, etc. Alguns livros: Varela, Thompson & Rosch, 1991; Edelman, 1992; Thelen & Smith, 1994; Port & van Gelder, 1995; Kelso, 1995; Clancey, 1997; Agre, 1997; Clark, 1997, 2001; Juarerro, 1999; Keijzer, 2001; e muitos outros Actually, this could be a good moment for a break or for some questions. Perception and action are not just stages. Infants are ‘unskilled’, adults are ‘skilled’. Action and perception are not just stages. CNS, body and world are always embedded and coupled. Development is the progressive ability to modulate this coupling Abstract mental activity is not only formed by, but stays dynamically coupled to perception and action. Perception and action are not just stages. Two examples: leg kicking and A-not-B error.

25 Sistemas incorporados e situados As ligações dinâmicas entre corpo, cérebro e mundo
coração A ligaçao dinamica entre mundo, cerebro e corpo Dedos dos pès: toes perceber, pensar decidir agir Os componentes para aperceber (waarnemen), pensar, decidir (beslissen) e agir, trocam (uitwisselen) mensagens, por meio de representações. It is not representational information exchange, it is direct dynamical coupling. CogSci 2002: Imagined action (Can I reach there with this rod) need not solved by thinking about rods and distances. Instead the rod becomes part of the body-brain system, thereby changing its intrinsic dynamics. This system, embedded in the world, generates an adaptive response. pés

26 Cognição incorporada Dinâmica intrínseca (Kelso, 1995)
Tendências de coordenação que são relativamente autônomas Aprender a subir em um local com grande inclinação (Adolph, 1993; Thelen & Smith, 1994) Aprender a dirigir Jogar futebol depois de muitos anos ….. Sistemas cognitivos se ajustam aos seus próprios corpos (Chiel & Beer, 1997) filogenético ontogenético the coordination tendencies are the hooks to hang on to for our thoughts and behavioral intentions and cognitive processes in general. Children spend a great deal of their first years learning their intrinsic dynamics. Did you ever try to play football after many years of not playing? I fell very often, because my legs were no longer carrying my weighty movements. Same after injuries (Willem’s trusting himself with tennis). Learning to drive a car. Cognitive systems ‘tune into their bodies’. Chiel & Beer (1997): evolution: walking upright led to changes (expansions) in vestibular system (larger anterior & posterior canals). Ontogenetically: changes in morphology leads to changes in e.g. sensomotoric maps. THM: the body has a mind of its own (Raibert & Hodgins, 1993). Cricket trachea in legs provide canals for additional phase & amplitude info Muscle tone is important for the actual effect of the neural signal Hydrostatic bodies (worms) allow different behavioral possibilities and impose diff constraints (e.g. in relation to force exertion) Info from self-induced movements is important for behavior. E.g. phasic feedback from stretch receptorsis important for maintaining normal flying movements in the locust. Vice versa bodies tune into their minds: if I stop doing sport, my body will change, if I increase my sport activities, my body will adapt too. Body important for valuing things. Not just doing it with your body, but also doing it for your body. Damasio’s Somatic marker hypothesis.

27 Cognição situada Andaimes
“podemos resolver, muitas vezes, problemas para utilizar propriedades confiáveis do ambiente. Este uso da estrutura externa é que entendo pela palavra ‘andaime’ (Clark, 1997, p.45)”. Sistemas cognitivos seguem ‘o principio do 007’ de James Bond: “Geralmente, criaturas evoluídas não armazenam nem processam informação de modo expendioso se eles podem usar a estrutura do ambiente e suas operações em conveniente substituição para operações de processamentos de informação. Ou seja, sabem só o que precisam saber para a realização do uma certa tarefa.” (Clark, 1997, p.46). "We manage our physical and spatial surroundings in ways that fundamentally alter the information-processing tasks our brains confront." (Clark, 1997, p.63). The 007 principle (Clark, 1997) "In general, evolved creatures will neither store nor process information in costly ways when they can use the structure of the environment and their operations upon it as a convenient stand-in for the information-processing operations. That is, know only as much as you need to know to get the job done." (Clark, 1997, p.46). Social scaffolding is important too. The point of epistemic actions is not to in the effect they have on the environment, but in the effect they have on the agent (K&M, p.546). Unearth info that is hard to compute. Perturb the world to force the re-evauluation of assumptions.

28 Cognição CIS implica uma conceptualização diferente da principal tarefa dum sistema cognitivo A tarefa não consiste em: focalizar o processo de resolução de problemas com base no processamento interno de informação Mas, ao invés disso: contribuir para a contínua interação do organismo com o meio ambiente Reconceptualize the tasks faced by cognitive systems: what it is doing, what it is for, why does it exist

29 O cérebro Não é um chefe de orquestra (Chiel & Beer, 1997)
Só um músico entre outros numa banda de jazz Não fundamentalmente orientado a resolver problemas Orientado ao improvisação Contribuindo a um repertório do comportamento O ambiente faz seleções do repertorio Chiel & Beer: Brain is not a conductor, but rather a player in a jazz ensemble. I suggest brain itself is a jazz ensemble that is improvising together with a big band. Also its contribution is guided by certain principles (lazy: not bebop but more like cool jazz).

30 Os termos primitivos das paradigmas na robótica
Input Output Sentido Dados sensoriais Informação percebida Plano Informação percebida e/ou cognitiva Diretrizes (instruções) Functions that take information from sensors and produce output used by other functions Functions take information (from sensors or internal knowledge) produce tasks to perform (go down the hall, turn left for 3 meters and stop). Functions produce output commands to motor actuators (turn 90 degrees clockwise with 0.2mps) Ação Informação percebida e/ou diretrizes Comandos aos efetores

31 O paradigma reativo Primitivos Input Output Sentido Dados sensoriais
Informação percebida Plano Informação percebida e/ou cognitiva Diretrizes (instruções) Functions that take information from sensors and produce output used by other functions Functions take information (from sensors or internal knowledge) produce tasks to perform (go down the hall, turn left for 3 meters and stop). Functions produce output commands to motor actuators (turn 90 degrees clockwise with 0.2mps) Ação Informação percebida e/ou diretrizes Comandos aos efetores

32 Decomposição vertical
Perceber criar modelo criar plano(s) raciocinar decidir fazer Decomposição horizontal Wander: andar distraidamente mapas explorar sensores efetores andar evitar

33 Criaturas reativas Estas criaturas não têm uma distinção clara entre componentes por percepção, intelecto central e ato E porque, não há um papel claro por representações As camadas competem entre si Toda camada é direcionada para transformar percepção no ato Toda camada constitua um padrão de interagem com o mundo (hábitos)

34 Comportamentos básicos e emergentes
Faz só uma coisa Estão independente (ligações fracas) e funcionam em paralelo Interagem por meio de combinação, supressão e inibição O comportamento ultimo é emergente Nenhum centro de controle Nenhum centro da inteligência Comportamento ‘parece inteligente’ Emergentie van ‘gewoon verschijnen’ naar ‘onverklaarbaar’. Hier unprogrammed functionality.

35 Plausibilidade biológica
Uso do conhecimento sobre animais Inteligência Cambriana maa (peixe – insetos) “the ability to move around in a dynamic environment sensing the surroundings to a degree sufficient to achieve the necessary maintenance of life and reproduction. This part of intelligence is where evolution has concentrated its time - it is much harder.” (Brooks, 1999, p.81). Perceber, detectar, observar Infrared sensors: seeking moving target (heat), 51 afm’s Cambrium mya *check insect evolution (boek Wilson over mieren bv). first fish 550 mya, insecten 450 mya, then things started moving fast. Reptiles 370, dinosaurs at 330, mammals 250, primates 120 great apes 18, “problem solving behavior, language, expert knowledge and application, and reason are all pretty simple once the essence of being and reacting,are available. That essence is [ zie hierboven] Genghis

36 Allen O primeiro mobot de Brooks 1984-1985 White line: just wandering
Blue line with additional layer for corridors (check)

37 O sojourner em Marte ‘Cool robot of the week’: Spirit (MER)
A Delta II rocket carrying the Spirit vehicle lifted clear of the Cape Canaveral Air Force Station in Florida at 1758 GMT on Tuesday. It was the third attempt at a blast-off - Sunday's and Monday's efforts were postponed because of stormy weather in the launch area. Spirit is identical to the Opportunity rover which will go into space later this month, probably on the 25th. The US rovers are going to Mars to study its geology, seeking chemical signatures that would confirm that abundant water once existed on the planet's surface. "The instrumentation on board these rovers, combined with their great mobility, will offer a totally new view of Mars, including a microscopic view inside rocks for the first time," Ed Weiler, Nasa headquarters associate administrator for space science, said earlier this week. The launch was successful"However, missions to Mars have proven to be far more hazardous than missions to other planets. "Historically, two out of three missions, from all countries who have tried to land on Mars, ended in failure. "We have done everything we can to ensure our rovers have the best chance of success," Dr Weiler said. Water clues Only three Nasa space craft have ever touched down successfully on Mars before: the Viking landers of 1976 and Mars Pathfinder. If all goes well, the rovers will head for two places that appear to have once been flooded by water. ROBOTS ON MARS What will the rovers do on Mars? Like Beagle 2, Nasa's rovers will bounce to airbag-cushioned landings at sites offering a balance of favourable conditions for safe landings and interesting science. Spirit is aimed at the Gusev Crater and Opportunity at a site called Meridiani Planum. "Gusev and Meridiani give us two different types of evidence about liquid water in Mars' history," said Dr Joy Crisp, Mer project scientist at the Jet Propulsion Laboratory in Pasadena. "Gusev appears to have been a crater lake. The channel of an ancient riverbed indicates water flowed right into it. "Meridiani has a large deposit of grey hematite, a mineral that usually forms in a wet environment." A new generation of robotic explorers are going to search for water on Mars. Mark Adler, deputy manager on the US space agency (Nasa) mission, talks to BBC News Online's Helen Briggs about what they might encounter. HB: How does each rover work? MA: It's a solar powered robot. At about 9 am it will wake up and by 2-3 pm in the afternoon it will send back what it has done. It then goes to sleep and our job begins here on Earth. We have about 19 hours to figure out the data and send a series of instructions to the rover next morning. HB: How similar are they to Pathfinder's Sojourner, the first Martian rover? MA: Pathfinder was quite similar in terms of its landing system for entry and descent to the surface of Mars. From that point on it looks a lot different to Pathfinder. The purpose of that mission overall was a technology demonstrator, but it did do some good science. Mars Exploration Rovers (Mer) is a science mission. Once the landers have delivered the rovers, they are discarded. Now the rovers are completely independent - they can go off to do work and have no reason to go back to the landers.

38 Conclusões O conhecimento comum deve ser reconsiderado à luz de uma perspectiva da cognição incorporada e situada: ao invés de descrições sobre o mundo, buscamos formas apropriadas de interação entre sujeito e ambiente. Regularmente nossos ações são baseados no processos reativos em vez de processos lógicos-dedutivos. Regularmente, nosso comportamento comum emerge da interação habitual do sujeito com mundo. A teoria da cognição incorporada e situada pode substituir a idéia da cognição lógico e puramente interna As teorias sistêmica e da auto-organização podem ajudar nossa compreensão do comportamento comum


Carregar ppt "Epistemologia e a Filosofia da Mente Incorporada e Situada"

Apresentações semelhantes


Anúncios Google