DreamCoder: growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning.

Expert problem-solving is driven by powerful languages for thinking about problems and their solutions. Acquiring expertise means learning these languages-systems of concepts, alongside the skills to use them. We present DreamCoder, a system that learns to solve problems by writing programs. It builds expertise by creating domain-specific programming languages for expressing domain concepts, together with neural networks to guide the search for programs within these languages. A 'wake-sleep' learning algorithm alternately extends the language with new symbolic abstractions and trains the neural network on imagined and replayed problems. DreamCoder solves both classic inductive programming tasks and creative tasks such as drawing pictures and building scenes. It rediscovers the basics of modern functional programming, vector algebra and classical physics, including Newton's and Coulomb's laws. Concepts are built compositionally from those learned earlier, yielding multilayered symbolic representations that are interpretable and transferrable to new tasks, while still growing scalably and flexibly with experience. This article is part of a discussion meeting issue 'Cognitive artificial intelligence'.

A language of thought for the mental representation of geometric shapes.

In various cultures and at all spatial scales, humans produce a rich complexity of geometric shapes such as lines, circles or spirals. Here, we propose that humans possess a language of thought for geometric shapes that can produce line drawings as recursive combinations of a minimal set of geometric primitives. We present a programming language, similar to Logo, that combines discrete numbers and continuous integration to form higher-level structures based on repetition, concatenation and embedding, and we show that the simplest programs in this language generate the fundamental geometric shapes observed in human cultures. On the perceptual side, we propose that shape perception in humans involves searching for the shortest program that correctly draws the image (program induction). A consequence of this framework is that the mental difficulty of remembering a shape should depend on its minimum description length (MDL) in the proposed language. In two experiments, we show that encoding and processing of geometric shapes is well predicted by MDL. Furthermore, our hypotheses predict additive laws for the psychological complexity of repeated, concatenated or embedded shapes, which we confirm experimentally.

Synthesizing theories of human language with Bayesian program induction.

Automated, data-driven construction and evaluation of scientific models and theories is a long-standing challenge in artificial intelligence. We present a framework for algorithmically synthesizing models of a basic part of human language: morpho-phonology, the system that builds word forms from sounds. We integrate Bayesian inference with program synthesis and representations inspired by linguistic theory and cognitive models of learning and discovery. Across 70 datasets from 58 diverse languages, our system synthesizes human-interpretable models for core aspects of each language's morpho-phonology, sometimes approaching models posited by human linguists. Joint inference across all 70 data sets automatically synthesizes a meta-model encoding interpretable cross-language typological tendencies. Finally, the same algorithm captures few-shot learning dynamics, acquiring new morphophonological rules from just one or a few examples. These results suggest routes to more powerful machine-enabled discovery of interpretable models in linguistics and other scientific domains.

Five Ways in Which Computational Modeling Can Help Advance Cognitive Science: Lessons From Artificial Grammar Learning.

There is a rich tradition of building computational models in cognitive science, but modeling, theoretical, and experimental research are not as tightly integrated as they could be. In this paper, we show that computational techniques-even simple ones that are straightforward to use-can greatly facilitate designing, implementing, and analyzing experiments, and generally help lift research to a new level. We focus on the domain of artificial grammar learning, and we give five concrete examples in this domain for (a) formalizing and clarifying theories, (b) generating stimuli, (c) visualization, (d) model selection, and (e) exploring the hypothesis space.

Factors influencing the caregiving performance of mothers of hospitalized toddlers with acute respiratory diseases: A path analysis.

Studies on the management of respiratory diseases in children have focused on family members' participation and caregivers' needs. However, evidence-based data on the effectiveness of mothers' management of acute respiratory diseases (ARDs) in toddlers are lacking. This study aimed to examine the factors influencing the caregiving performance of mothers of toddlers hospitalized for an ARD and to test a hypothetical causal model based on the Caregiving Effectiveness Model (CEM). A cross-sectional design was used, and participants included 291 mothers of toddlers aged 12-36 months who were hospitalized for an ARD. Based on the CEM, data were analyzed to identify the path of relationships between the factors influencing mothers' care of their hospitalized children and the mothers' caregiving performance. The modified path model had a good fit with the data, with optimal values for all fit indices. The mothers' caregiving performance was influenced by the children's number of hospitalizations, the mother-child relationship, and the mothers' anxiety level. These three factors explained 51.4% of the variance in the mothers' caregiving performance. Educational interventions targeting controllable factors such as mother-child relationships and mothers' anxiety levels may be considered to improve mothers' caregiving performance.

Development and Preliminary Testing of a Smartphone-Based Injury-Prevention Application (S-IPA) for Teachers at Child-Care Centers in South Korea.

The purpose of this study was to develop a smartphone-based injury-prevention application (S-IPA) for teachers working in child-care centers, and to test the satisfaction level of the users of the application (app). Through a literature review and needs assessment, an app compatible with the Apple iPhone operating system was developed. The app was verified and the mean total satisfaction with 7 features of the app was 7.76 (± 1.13) on a score of 1-10. The result of the S-IPA survey showed a positive response, indicating a high potential for use as a teacher's educational guide, which would provide an effective information delivery system for the prevention of possible injuries at child-care centers.

Muscle histopathology in children with spastic cerebral palsy receiving botulinum toxin type A.

INTRODUCTION: Botulinum toxin A (BoNTA) is routine treatment for hypertonicity in children with cerebral palsy (CP). METHODS: This single-blind, prospective, cross-sectional study of 10 participants (mean age 11 years 7 months) was done to determine the relationship between muscle histopathology and BoNTA in treated medial gastrocnemius muscle of children with CP. Open muscle biopsies were taken from medial gastrocnemius muscle and vastus lateralis (control) during orthopedic surgery. RESULTS: Neurogenic atrophy in the medial gastrocnemius was seen in 6 participants between 4 months and 3 years post-BoNTA. Type 1 fiber loss with type 2 fiber predominance was significantly related to the number of BoNTA injections (r = 0.89, P < 0.001). CONCLUSIONS: The impact of these changes in muscle morphology on muscle function in CP is not clear. It is important to consider rotating muscle selection or injection sites within the muscle or allowing longer time between injections.

Autologous chondrocyte implantation of the ankle: a 2- to 5-year follow-up.

BACKGROUND: Treatment of full-thickness talar cartilage defects that have failed previous surgery is problematic without a definitive solution. PURPOSE: To report the first US prospective study of autologous chondrocyte implantation of the talus. STUDY DESIGN: Case series; Level of evidence, 4. METHODS: Eleven patients (6 women and 5 men; mean age, 33 years) underwent autologous chondrocyte implantation of the talus after previous failed surgical management. There were 9 medial and 2 lateral lesions, with a mean size of 21 x 13 mm (273 mm2). Five patients underwent autologous chondrocyte implantation of the talus alone; 6 had it with a "sandwich procedure." Ten patients underwent a second-look arthroscopy with screw removal. RESULTS: Mean follow-up was 38 months. Preoperatively, 10 patients rated their ankles as poor and 1 as fair, using the simplified symptomatology evaluation. At latest follow-up, 3 patients were classified as excellent, 6 as good, and 2 as fair. Tegner activity level improved from 1.3 +/- 1.0 (mean +/- SE) preoperatively to 4.0 +/- 1.6 (P < .002) postoperatively. The Finsen score (modified Weber score) showed significant improvement in the total score (P < .001). There was also overall agreement between the Finsen score and the American Orthopaedic Foot and Ankle Society ankle hindfoot score, with significant improvement from 47.4 +/- 17.4 preoperatively to 84.3 +/- 8.1 postoperatively (P < .001). At repeat arthroscopy, complete coverage of the defect was seen in all patients. CONCLUSION: Autologous chondrocyte implantation of the talus yields significant functional improvement; however, further investigation is necessary to determine the long-term structural and biomechanical properties of the repair tissue.