a computational pragmatics for weaseling
play

A Computational Pragmatics for Weaseling An implementation in the - PowerPoint PPT Presentation

A Computational Pragmatics for Weaseling An implementation in the RSA-framework Leander Vignero KU Leuven, Institute of Philosophy (CLPS) April 12, 2019 What is weaseling? And what are probabilistic expressions? Oxford English Dictionary:


  1. A Computational Pragmatics for Weaseling An implementation in the RSA-framework Leander Vignero KU Leuven, Institute of Philosophy (CLPS) April 12, 2019

  2. What is weaseling? And what are probabilistic expressions? Oxford English Dictionary: Achieve something by use of cunning or deceit. 1 North American Behave or talk evasively. 2 simple, technical definition: misrepresenting one’s degrees of belief/ beliefs through vague verbiage. What are PEs? Examples: probably , unlikely , about even , possibly , . . . 1 Weaseling and PEs in RSA

  3. Life Hack: Avoiding Meetings meeting enthusiast: Are you coming to the meeting tomorrow? You: . . . 2 Weaseling and PEs in RSA

  4. Tell the truth meeting enthusiast: Are you coming to the meeting tomorrow? You: No, I won’t come. Downside: they are quite likely to ask a follow-up question: “Oh, why not?” 3 Weaseling and PEs in RSA

  5. Lie outright meeting enthusiast: Are you coming to the meeting tomorrow? You: Yes, I will come. Downside: You break your word. (Great social cost) 4 Weaseling and PEs in RSA

  6. Weasel out from under the situation meeting enthusiast: Are you coming to the meeting tomorrow? You: Yes, I might come./ I will try to come. / Maybe. Probabilistic expressions lower your accountability! 5 Weaseling and PEs in RSA

  7. Another example Tetlock’s example (Tetlock et al. , 2017): A forecaster says there is a distinct possibility of p happening. If p happens the forecaster can say: I told you it could happen. 1 If p does not happen, they can say: It was a mere possibility. 2 The study of probabilistic expressions originated in the intelligence studies (Kent, 1964). 6 Weaseling and PEs in RSA

  8. Why should you care? Academic Interest per se 1 Modeling/understanding human languages 2 Relation to other issues in social epistemology (specifically testimony) 3 7 Weaseling and PEs in RSA

  9. Some housekeeping I will tell you about the PE-literature and the RSA-literature 1 I will tell you how we can use these ideas to create an idealized language 2 to think about communicating uncertainty and about weaseling 8 Weaseling and PEs in RSA

  10. What are PEs? Probabilistic expressions are used to communicate uncertainty (Kent 1964): 9 Weaseling and PEs in RSA

  11. Thinking about these problems It’s all nice and easy to speculate and philosophize about these things. But it would be even better to have an idealized language to analyze these communicative situations (communication of uncertainty and weaseling). This is a reinterpretation of the RSA-framework: using it as an idealized prescriptive framework, rather than a descriptive framework. Why not probabilistic dynamic epistemic logic (probabilistic DEL)? probabilistic DEL is too strong: there is no real room for vagueness. 1 probabilistic DEL does not have a “critical” theory of mind. 2 10 Weaseling and PEs in RSA

  12. The Rational Speech Act Framework (RSA) Framework from computational cognitive psychology that operationalizes the Gricean idea of cooperation probabilistically. Advantages: Limited set of utterances. 1 Theory of mind. 2 Picks up on implicatures. 3 11 Weaseling and PEs in RSA

  13. Philosophy of language crash course: Gricean pragmatics Grice thought of language as a cooperative venture: speakers choose their words in such a way that they are best understood. Listeners, in turn, know this and use this information to further narrow the scope of interpretation. RSA turns “in such a way that they are best understood” into “in a way that maximizes the probability of being correctly understood”. 12 Weaseling and PEs in RSA

  14. Scalar implicature “The idea is that if the speaker were in a position to make the stronger statement, he would have. Since he did not, he must believe that the stronger statement is not true.” – Wayne Davis Modeled by Goodman and Stuhlm¨ uller (2013) using an RSA-style model for utterances like none , some and all . For the philosophers and modal logicians: this also holds for classic modal notions like possibility , necessity . 13 Weaseling and PEs in RSA

  15. The literal listener RSA is a layered model that produces degrees of belief regarding states of the world upon hearing an utterance (Frank & Goodman , 2012). A literal listener that interprets utterances literally: L 0 ( s | u ) ∝ P ( s ) χ � u � ( s ) . (1) In words: the literal listener interprets utterances u in proportion to their priors with respect to the states s that belong to the meaning of u . This amounts to Bayesian updating. 14 Weaseling and PEs in RSA

  16. The literal listener Consider the utterances: some and all . (a) Prior (b) L 0 ( · | some ) 15 Weaseling and PEs in RSA

  17. The pragmatic speaker Utility function that determines how well an utterance succeeds at communicative aims: U ( u | s ) = − log( L 0 ( s | u )) . (2) A (subrational) pragmatic speaker: S 1 ( u | s ) ∝ exp( λ U ( u | s )) . (3) The pragmatic speaker picks words proportionally to the power of the utilities. 16 Weaseling and PEs in RSA

  18. The pragmatic speaker (a) U ( · | 6) (b) S 1 ( · | 6) 17 Weaseling and PEs in RSA

  19. The pragmatic listener A pragmatic listener: L 1 ( s | u ) ∝ P ( s ) S 1 ( s | u ) . (4) Figure: L 1 ( · | some ) 18 Weaseling and PEs in RSA

  20. Quick Recap RSA is a layered model that uses Bayesian inference to infer knowledge states of agents that are situated lower in the hierarchy. For instance: the pragmatic speaker S 1 is defined in terms of the literal listener L 0 (mediated by the utility function). This is why this framework has such an interesting theory of mind. 19 Weaseling and PEs in RSA

  21. Some more housekeeping The work by Yoon et al. (2016, 2017, 2018) on politeness is a starting point. a variation on Yoon et al. (2016, 2017, 2018) 1 a variation on Bergen and Goodman (2015) 2 a variation on Yoon et al. (2016, 2017, 2018) 3 a personal variation 4 I will start with two honest speakers and then move to weasels. I will conclude with one slide on the implementation in Julia. 20 Weaseling and PEs in RSA

  22. Hearing the nuance of PEs Think about the states as probabilities ( 0 ; 0 . 2 ; 0 . 4 ; 0 . 6 ; 0 . 8 , 1 ). We need to adjust equation 1 from L 0 ( s | u ) ∝ P ( s ) χ � u � ( s ) to: L 0 ( s | u ) ∝ P ( s ) N ( u )( s ) . (5) (Variation on Yoon et al. (2016, 2017, 2018). (a) Normal. (b) With nuance. 21 Weaseling and PEs in RSA

  23. Communicating a distribution We need to update both the listeners and the utility function. We need to assume a set of probability distributions that can be communicated. For now let us assume they are the posteriors of observations . L 0 ( s, o | u ) ∝ P ( s ) P ( o | s ) N ( u )( s ) (6) What really matters is the utility function: U ( u | o ) = − KL D ( P ( · | o ) || L 0 ( · | u )) (7) (Variation on Bergen and Goodman (2015)) 22 Weaseling and PEs in RSA

  24. Two small problems Are we always communicating posteriors of well-defined observations? 1 Is there always common belief about the distributions that could be 2 communicated? Two solutions Finding ways of modeling the communication of other types of 1 distributions. An alternative I propose later on. 2 23 Weaseling and PEs in RSA

  25. Tweeking the utility function Introduce a new utility function U ego , that only concerns non-epistemic utilities. Then the total utility, U total , is a convex combination of U epistemic and U ego : U total = β U epistemic + (1 − β ) U ego . (8) β represents the honesty of the speaker. (idea from Yoon et al. (2016, 2017, 2018)) With such utility functions we can try to understand weaseling! Speakers can try to weasel out from under situations, whilst listeners can try to catch them in the act! 24 Weaseling and PEs in RSA

  26. An alternative A four-layered model: L 0 ( s | u ) ∝ P ( s ) N ( u )( s ) (9) S 1 ( u | s ) ∝ exp( λ U ) (10) L 1 ( s | u ) ∝ P ( s ) S 1 ( u | s ) (11) S 2 ( u ) ∝ exp( λKL D ( P ( · ) || L ( · | u ))) (12) 25 Weaseling and PEs in RSA

  27. Implementation in the Julia programming language procedural implementation. 1 The crux is that we can represent probability distributions as unit vectors. 2 26 Weaseling and PEs in RSA

  28. Conclusion/Upshot I have provided a provisional philosophical analyis of weaseling. 1 I have created a couple of computational models for thinking about 2 communicating uncertainty with PEs. I have created computational models for thinking about weaseling. 3 I have implemented these frameworks in the Julia programming language. 4 27 Weaseling and PEs in RSA

  29. References Bergen, L. and Goodman, N. D. (2015). The strategic use of noise in pragmatic reasoning. Topics in cognitive science , 7(2):336–350. 28 Weaseling and PEs in RSA

  30. References Frank, M. C. and Goodman, N. D. (2012). Predicting pragmatic reasoning in language games. Science , 336(6084):998–998. Goodman, N. D. and Stuhlm¨ uller, A. (2013). Knowledge and implicature: Modeling language understanding as social cognition. Topics in cognitive science , 5(1):173–184. 29 Weaseling and PEs in RSA

  31. References Kent, S. (1964). Words of estimated probability. Intelligence Studies , (8):49–65. Tetlock, P. E., Mellers, B. A., and Scoblic, J. P. (2017). Bringing probability judgments into policy debates via forecasting tournaments. Science , 355(6324):481–483. 30 Weaseling and PEs in RSA

Recommend


More recommend