Probing the compositionality of intuitive functions
How do people learn about complex functional structure? Taking inspiration from other areas of cognitive science, we propose that this is accomplished by harnessing compositionality: complex structure is decomposed into simpler building blocks. We formalize this idea within the framework of Bayesian...
| Main Authors: | , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Published: |
Neural Information Processing Systems Foundation
2017
|
| Online Access: | http://hdl.handle.net/1721.1/112750 https://orcid.org/0000-0002-1925-2035 |
| _version_ | 1811071990649323520 |
|---|---|
| author | Schulz, Eric Duvenaud, David K. Speekenbrink, Maarten Gershman, Samuel J. Tenenbaum, Joshua B |
| author2 | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences |
| author_facet | Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Schulz, Eric Duvenaud, David K. Speekenbrink, Maarten Gershman, Samuel J. Tenenbaum, Joshua B |
| author_sort | Schulz, Eric |
| collection | MIT |
| description | How do people learn about complex functional structure? Taking inspiration from other areas of cognitive science, we propose that this is accomplished by harnessing compositionality: complex structure is decomposed into simpler building blocks. We formalize this idea within the framework of Bayesian regression using a grammar over Gaussian process kernels. We show that participants prefer compositional over non-compositional function extrapolations, that samples from the human prior over functions are best described by a compositional model, and that people perceive compositional functions as more predictable than their non-compositional but otherwise similar counterparts. We argue that the compositional nature of intuitive
functions is consistent with broad principles of human cognition. |
| first_indexed | 2024-09-23T08:59:10Z |
| format | Article |
| id | mit-1721.1/112750 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T08:59:10Z |
| publishDate | 2017 |
| publisher | Neural Information Processing Systems Foundation |
| record_format | dspace |
| spelling | mit-1721.1/1127502022-09-26T09:39:11Z Probing the compositionality of intuitive functions Schulz, Eric Duvenaud, David K. Speekenbrink, Maarten Gershman, Samuel J. Tenenbaum, Joshua B Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences Tenenbaum, Joshua B How do people learn about complex functional structure? Taking inspiration from other areas of cognitive science, we propose that this is accomplished by harnessing compositionality: complex structure is decomposed into simpler building blocks. We formalize this idea within the framework of Bayesian regression using a grammar over Gaussian process kernels. We show that participants prefer compositional over non-compositional function extrapolations, that samples from the human prior over functions are best described by a compositional model, and that people perceive compositional functions as more predictable than their non-compositional but otherwise similar counterparts. We argue that the compositional nature of intuitive functions is consistent with broad principles of human cognition. 2017-12-14T15:10:43Z 2017-12-14T15:10:43Z 2016-12 2017-12-08T13:14:34Z Article http://purl.org/eprint/type/JournalArticle http://hdl.handle.net/1721.1/112750 Schulz, Eric et al. "Probing the Compositionality of Intuitive Functions." Advances in Neural Information Processing Systems 29 (NIPS 2016), Barcelona, Spain, December 5-10, 2016. © 2016 Neural Information Processing Systems Foundation https://orcid.org/0000-0002-1925-2035 https://papers.nips.cc/paper/6130-probing-the-compositionality-of-intuitive-functions 30th Conference on Neural Information Processing Systems (NIPS 2016) Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Neural Information Processing Systems Foundation Neural Information Processing Systems (NIPS) |
| spellingShingle | Schulz, Eric Duvenaud, David K. Speekenbrink, Maarten Gershman, Samuel J. Tenenbaum, Joshua B Probing the compositionality of intuitive functions |
| title | Probing the compositionality of intuitive functions |
| title_full | Probing the compositionality of intuitive functions |
| title_fullStr | Probing the compositionality of intuitive functions |
| title_full_unstemmed | Probing the compositionality of intuitive functions |
| title_short | Probing the compositionality of intuitive functions |
| title_sort | probing the compositionality of intuitive functions |
| url | http://hdl.handle.net/1721.1/112750 https://orcid.org/0000-0002-1925-2035 |
| work_keys_str_mv | AT schulzeric probingthecompositionalityofintuitivefunctions AT duvenauddavidk probingthecompositionalityofintuitivefunctions AT speekenbrinkmaarten probingthecompositionalityofintuitivefunctions AT gershmansamuelj probingthecompositionalityofintuitivefunctions AT tenenbaumjoshuab probingthecompositionalityofintuitivefunctions |