Analogies Chapter 5

1. AnalogiesChapter 5 Paul Thagard (2005). Mind: An Introduction to Cognitive Science. 2nd Edition. MIT Press FCAC, University of Hyderabad

2. Outline • Analogies as Mental Representations • Evaluation of Analogies as a Representation scheme • Representational power • Computational power • Problem Solving: Planning, Decision, Explanation • Learning • Language • Psychological Plausibility • Neurological Plausibility • Practical Applicability FCAC, University of Hyderabad

3. Introduction • Analogical thinking consists of dealing with a new situation by adapting a similar situation. • In concepts you induce a general rule/concept from particular situations. • Even if the particular experiences are too limited to formulate a rule, you can use analogies to formulate behavior. • Human use of analogy is very old and … • All the epics like Mahabharata, Ramayana, Bible, Iliad used analogies to guide behavior in the reader’s own situation. • Continues still … • Recent one is, Gurucharan Das (2009): The Difficulty of Being Good: On the Subtle Art of Dharma, rules of conduct for society, business based on analogies to Mahabharata. • In AI, Analogies = Case Based Reasoning (CBR) FCAC, University of Hyderabad

4. Introduction … • Verbal and visual representations of situations that can be used as analogs or cases. • Processes of retrieval, mapping, and adaptation operate on those analogs • The analogical processes, applied to the representations of analogs, produce intelligent behaviour. • The constraints of similarity, structure, and purpose guide the process of how previous experiences can be found and used to help with new problems. FCAC, University of Hyderabad

5. Representational Power • Analogy is a systematic relationship between two situations, the target analog(ue) representing the new situation to be reasoned about and the source analog(ue) representing the old situation that can be adapted and applied to the target analog. • Representing of analogies includes identifying surface similarities or features and also representing causal relations that produced the outcomes relevant to your goals. • Example: registering for an elective based on previous experience • Analogs are like concepts, rules in terms of packaging of information but unlike these, analogy is about particular situation. • Visual representation of analogs (as in images). • Ex: Using the mental picture of familiar building to get around in an unfamiliar one. FCAC, University of Hyderabad

6. Computational Power • Analogical reasoning is useful when you have some experience with a domain but little general knowledge • Computationally powerful when conceptual and rule-based knowledge is not available. • In AI, such schemes are called instance based (learning) methods. Other examples: kNN, RBF networks • Computational procedure: • Remembering (retrieving from memory) • Comparing: mapping the source and target analogs to each other • Adaptation FCAC, University of Hyderabad

7. Computational Power • Combinatorial Explosion: • How to retrieve cases from the case-base? At 10 tasks/day for the past 20 years, you have potentially stored 73,000 task solutions!! • Factors important for retrieval • Retrieval governed by three constraints: similarity, structure, and purpose (Holyoak and Thagard, 1995). • Forbus, Gentner, and Law (1995) argue only similarity important, not the other two. • Creative analogies involve leaps in thinking. • The juvenile sea squirt wanders through the sea searching for a suitable rock or hunk of coral to cling to and its home for life. For this task, it has a rudimentary nervous system, When it finds its spot and takes root, it doesn’t need its brain anymore, so it eats it! (It’s rather like getting tenure.) Dennett, 1991. • How is a professor getting tenure like a sea squirt eating its brain? FCAC, University of Hyderabad

8. Computational Power (contd.) • Planning: How to accomplish a goal from an initial or current state? • Retrieve from memory a case that matches the situation and goals. • Ex: Solving exercises at the end of the chapter by looking at solved examples in the main text. • If the target problem is genuinely novel, case-based reasoning is problematic. FCAC, University of Hyderabad

9. Computational Power (contd.) • Decision: Selection of best means to get to the goal. • Similar to planning, decisions based on familiar cases • Legal reasoning often involves reference to previous cases that serve as precedents. • US debate on Saddam Hussein’s (SH) invasion of Kuwait: SH : Adolf Hitler, there fore KI : WWII invasion of Germany • Critics drew analogies to US’s disastrous involvement in Vietnam • Often problem is when you are fixated on a single analogy and ignore other relevant source analogs that offer different actions to choose from. FCAC, University of Hyderabad

10. Computational Power (contd.) • Explanation: In this, you are trying to understand why something happened. • Analogies important solutions to the explanation problem in education and research situations. • Teachers often explain unfamiliar things by comparison with what the students already know. • Ex: Baseball in US is similar to Indian Cricket, highlighting similarities and differences • Analogical explanations are often limited by the fact that things being compared have many differences as well as similarities. • Fundamental analogy in CogSci: Mind : Computer!! FCAC, University of Hyderabad

11. Computational Power (contd.) • Learning: Ability to use experience to improve performance. • Analogical thinking involves three kinds of learning • (1) The storage of cases based on previous experience, storing instances, low-level, instance-based learning. • (2) Directly as a result of analogizing, when you adapt a previous case to solve a new problem (including abductive process of analogizing). • Ex: If your instructor is late for class, you may remember a previous case when he was late because of flat tire, car problem, emergency, etc. • (3) Introducing a general element. Forming an analogical schema that captures what is common to both the source and target analogs. • Analogical schema are different from conceptual schemas in that they are not as general and are usually based on only one or two instances! FCAC, University of Hyderabad

12. Computational Power (contd.) • Language: How do concepts underlie our ability to use language? • Analogy plays an important role in the production and comprehension of language, since it underlies metaphor. • Ex: Bengalure is the Silicon Valley of India. • Ex: Marriage is a battlefield. Source (war) and Target (marriage)! • Ex: Information Superhighway is like a fast and efficient highway for electronic data. • Some people think metaphor as a deviant use of language. Many people see it as a pervasive and valuable feature of language. • Metaphors may go beyond analogy by using other figurative devices to produce a broader aura of associations. Lakoff is a prominent researcher in metaphors. FCAC, University of Hyderabad

13. Psychological Plausibility • Psychological plausibility assessed through experiments on how people use analogies • Example: In the Tumor Problem, most people find it hard to think how the doctor can use the rays to kill the tumor without destroying the healthy tissue. FCAC, University of Hyderabad

14. Tumor Problem Suppose you are a doctor faced with a patient who has a malignant tumor in his stomach. It is impossible to operate on the patient, but unless the tumor is destroyed the patient will die. There is a kind of ray that can be used to destroy the tumor. If the rays reach the tumor all at once at a sufficiently high intensity, the tumor will be destroyed. Unfortunately, at this intensity the healthy tissue that the rays pass through on the way to the tumor will also be destroyed. At lower intensities the rays are harmless to healthy tissue, but they will not affect the tumor either. What type of procedure might be used to destroy the tumor with the rays and at the same time avoid destroying the healthy tissue? Glick and Holyoak, 1980 FCAC, University of Hyderabad

15. Psychological Plausibility • In the Tumor Problem, Glick and Holyoak (1980) found that only 10 percent of college students could produce a good solution. • In contrast, 75 percent of college students produce a good solution if they were told the fortress story. FCAC, University of Hyderabad

16. Fortress Story A small country fell under the iron rule of a dictator. The dictator ruled the country from a strong fortress. The fortress was situated in the middle of the country, surrounded by farms and villages. Many roads radiated outward from the fortress like spokes on a wheel. A great general arose who raised a large army at the border and vowed to capture the fortress and free the country of the dictator. The general knew that if entire army could attack the fortress at once it could be capture. His troops were poised at the head of one of the roads leading to the fortress, ready to attack. However, a spy brought the general a disturbing report. The ruthless dictator had planted mines on each of the roads. The mines were set so that small bodies of men could pass over them safely, since the dictator needed to be able to move troops and workers to and from the fortress. However, any large force would detonate the mines. Not only would this blow up the road and render it impassable, but the dictator would destroy many villages in retaliation. A full-scale direct attack on the fortress therefore appeared impossible. The general, however, was undaunted. He divided his army up into small groups and dispatched each group to the head of different road. When all was ready he gave the signal, and each group charged down a different road. All of the small groups passed safely over the mines, and the army then attacked the fortress in full strength. In this way, the general was able to capture the fortress and overthrow the dictator. Glick and Holyoak, 1980 FCAC, University of Hyderabad

17. Psychological Plausibility • Shows how analogical learning helps solve new problems. • Glick and Holyoak also investigated how multiple analogical schema can help problems solving. Apart from Fortress story, they gave a story of a firefighter who extinguished oil-well fire by using multiple small hoses. FCAC, University of Hyderabad

18. Psychological Plausibility • Behavioral experiments on interference between Metaphorical versus Literal meaning • Ex. Response time measured when students have to decide “Some desks are junkyards” ( literally false but metaphorically true). Versus “Some Desks are roads”. (Literally and metaphorically false). People are slower on the former as compared to the latter. • Metaphorical interpretation appears to be an obligatory process that accompanies literal processing, rather than being an optional process that occurs after literal processing. FCAC, University of Hyderabad

19. Neurological Plausibility • Neurological research on analogical reasoning is in its infancy. • rTMS (reversible transcranial magnetic stimulation) expts show left prefrontal cortex involved in analogical reasoning. • Neural network models have been constructed to demonstrate analogical reasoning. FCAC, University of Hyderabad

20. Practical Applicability • Analogies are very important in education • Effective teachers use analogies to make unfamiliar material easy to comprehend. • Holyoak and Thagard (1995) gave recommendations for how educators can successfully use analogies. • Use familiar sources, make the mapping clear, use deep, systematic analogies, describe the mismatches, use multiple analogies, perform analogy therapy • Engineering design • Analogies are fertile sources for creative designs. Ex: Velcro; Telephone based on human ear; reverse engineering. HCI – desktop in windows and Mac. • Case-based Reasoning systems used extensively in the industry for diagnosis, design, etc. FCAC, University of Hyderabad

21. References • Paul Thagard (2005). Mind: An Introduction to Cognitive Science. 2nd Edition. MIT Press. FCAC, University of Hyderabad