Learning to learn causal models

Learning to understand a single causal system can be an achievement, but humans must learn about multiple causal systems over the course of a lifetime. We present a hierarchical Bayesian framework that helps to explain how learning about several causal systems can accelerate learning about systems that are subsequently encountered. Given experience with a set of objects, our framework learns a causal model for each object and a causal schema that captures commonalities among these causal models. The schema organizes the objects into categories and specifies the causal powers and characteristic features of these categories and the characteristic causal interactions between categories. A schema of this kind allows causal models for subsequent objects to be rapidly learned, and we explore this accelerated learning in four experiments. Our results confirm that humans learn rapidly about the causal powers of novel objects, and we show that our framework accounts better for our data than alternative models of causal learning.     

Learning to learn: From within-modality to cross-modality transfer during infancy

One critical aspect of learning is the ability to apply learned knowledge to new situations. This ability to transfer is often limited, and its development is not well understood. The current research investigated the development of transfer between 8 and 16 months of age. In Experiment 1, 8- and 16-month-olds (who were established to have a preference to the beginning of a visual sequence) were trained to attend to the end of a sequence. They were then tested on novel visual sequences. Results indicated transfer of learning, with both groups changing baseline preferences as a result of training. In Experiment 2, participants were trained to attend to the end of a visual sequence and were then tested on an auditory sequence. Unlike Experiment 1, only older participants exhibited transfer of learning by changing baseline preferences. These findings suggest that the generalization of learning becomes broader with development, with transfer across modalities developing later than transfer within a modality.     

Learning how to learn at university: The case for an experimental centre

It is proposed that an experimental 'learning how to study centre' should be set up in universities. Reduction of wastage and attrition is likely to be brought, about not by more refined selection procedures but by increased skill in tuition amongst university teachers and by increased skill in self-tuition amongst students. The proposed centre would be manned by competent staff and would provide technical facilities for practical exploration of the problems of tuition and learning for those tutors and students who wish to improve their tuition and study skills. In particular, it would explore the extent to which learning skills can be improved through active self-tuition and through the activity of 'learning through teaching'.     

Learning to learn during visual discrimination in group housed dwarf goats (Capra hircus)

Using an automated learning device, we investigated "learning to learn" by dwarf goats (Capra hircus) in what was for them a familiar environment and normal social settings. Nine problems, each consisting of four discriminable black symbols, each with one S-super+ and three different S-super(-), were presented on a computer screen. Mean daily learning success improved over the course of the first four problems, and the improvement was maintained throughout the remaining five problems. The number of trials to reach the learning criterion decreased significantly beginning with problem four. Such results may be interpreted as evidence that the goats were developing a learning set. In the present case, the learning set appeared to have two components. One involved gaining familiarity and apparent understanding of the learning device and the basic requirements of the discrimination task. The second component involved learning potential error factors to be ignored, as well as learning commonalities that carried over from one problem to the next. Among the error factors, evidence of apparent preferences for specific symbols was seen, which had a predictable effect on performances.     

Biased to learn language

Some recent publications that explore the foundations of early language development are reviewed in this article. The review adopts the pivotal idea that infants' advancements are helped by the existence of different types of biases. The infant's discovery of the phonological properties of the language of the environment, as well as their learning of words, is assisted by different types of processing biases, ranging from perceptual to social factors.     

Learning the value of money from stochastically varying prices

In 3 experiments, the authors investigated learning of the value of money from product prices in an unfamiliar currency when the prices are proportional to quantity. In support of the second stage of a hypothesized 2-stage process of learning, Experiment 1, in which 32 undergraduates participated, shows that response times for inferences of quantity are longer when participants are presented with quantity-price pairs than when they are presented with price-quantity pairs. Experiments 2 and 3, in which 54 and 34 undergraduates participated, respectively, show that (a) stochastic price variation causes systematic errors in the learning of unit prices from quantity-price pairs as a result of judgmental regression effects and (b) in support of the 2-stage learning hypothesis, inferences of quantity are the inverse of the learned unit prices.     

Learning your way around town: how virtual taxicab drivers learn to use both layout and landmark information

By having subjects drive a virtual taxicab through a computer-rendered town, we examined how landmark and layout information interact during spatial navigation. Subject-drivers searched for passengers, and then attempted to take the most efficient route to the requested destinations (one of several target stores). Experiment 1 demonstrated that subjects rapidly learn to find direct paths from random pickup locations to target stores. Experiment 2 varied the degree to which landmark and layout cues were preserved across two successively learned towns. When spatial layout was preserved, transfer was low if only target stores were altered, and high if both target stores and surrounding buildings were altered, even though in the latter case all local views were changed. This suggests that subjects can rapidly acquire a survey representation based on the spatial layout of the town and independent of local views, but that subjects will rely on local views when present, and are harmed when associations between previously learned landmarks are disrupted. We propose that spatial navigation reflects a hierarchical system in which either layout or landmark information is sufficient for orienting and wayfinding; however, when these types of cues conflict, landmarks are preferentially used.     

Form follows function: Learning about function helps children learn about shape

Object functions help young children to organize new artifact categories. However, the scope of their influence is unknown. We explore whether functions highlight property dimensions that are relevant to artifact categories in general. Specifically, using a longitudinal training procedure, we assessed whether experience with functions highlights shape as important for categorization. Seventeen-month-olds were provided experience with novel categories of similarly shaped objects. In Study 1, the function group learned about the objects’ shape-based functions; a control group did not. In Study 2, 17-month-olds were trained on the same categories, but the objects’ shapes and functions were no longer causally related. Only the function group in Study 1 subsequently used shape reliably when categorizing novel objects. These results suggest that function is instrumental in establishing a ‘shape bias’ in early categorization and that it does so via conceptually based processes.     

How and when infants learn to climb stairs

Seven hundred and thirty-two parents reported when and how their infants learned to climb stairs. Children typically mastered stair ascent (mean age=10.97 months) several months after crawling onset and several weeks prior to descent (mean age=12.53 months). Most infants (94%) crawled upstairs the first time they ascended independently. Most infants (76%) turned around and backed at initial descent. Other descent strategies included scooting down sitting, walking, and sliding down face first. Children with stairs in their home were more likely to learn to ascend stairs at a younger age, devise backing as a descent strategy, and be explicitly taught to descend by their parents than children without stairs in their home. However, all infants learned to descend stairs at the same age, regardless of the presence of stairs in their home. Parents' teaching strategies and infants' access to stairs worked together to constrain development and to influence the acquisition of stair climbing milestones.     

The bizarre imagery effect and intention to learn

The bizarre imagery effect, better memory for bizarre stimuli than for common stimuli, is now an established finding. However, the mnemonic benefits of bizarre imagery are subject to several constraints (e.g., the use of mixed lists and free-recall tests). A further constraint on the bizarreness effect is demonstrated here. In each of two experiments, subjects were given either incidental or intentional study instructions and were asked to rate the vividness of the images they formed from the bizarre and common sentences. Contrary to conclusions based on available evidence, the bizarreness effect in free recall was manifested only with the incidental learning instructions. This additional constraint on the effect is consistent with the item-order account of bizarreness.     

How toddlers begin to learn verbs

Any theory of language must account for how children learn verbs, the gateway to grammar. Yet verbs can be difficult to learn. Building on Gentner's 'natural partitions hypothesis' we suggest that, to learn a verb, infants must conceptualize components of events and map verbs in the ambient language onto those components. Although toddlers detect and categorize at least some of the conceptual underpinnings of verb categories, the mapping of verbs onto these representations is not transparent. Mapping is a difficult problem in its own right. The Emergentist Coalition Model that has been used to explain noun learning also begins to explain how children move from perceptual to social and then to linguistic information to link verbs to actions and events.     

Rats learn to eat more to avoid hunger

Several recent experiments have provided evidence that the ingestion of a distinctive food by rats can be a learnt instrumental act as well as an associatively conditioned reaction. In the previous work, maintenance food was withheld for shorter and longer durations on different days following access to the training food. Extra eating before the longer fast was interpreted as avoidance of hunger. This interpretation was based on the evidence showing that extra eating as a result of classical conditioning comes from pairing food stimuli with the presence of little or no hunger because of repletion with energy nutrients. The theory that the extra eating arose from a response-depletion contingency was tested in the present experiment by training rats on only a long fast or only a short fast. Greater increase in intake was seen before the longer fast. The results also replicated previously seen cycles of increase, decrease, and renewed increase in putative deficit-avoidant eating over about three trials, indicating that the extra eating reduces the response-reinforcing hunger and that the consequent part-extinction restores reinforcement. The shape of the learning curve was consistent with these cycles occurring from the start of training, further supporting the view that the increase in food intake before a long delay in refeeding is hunger-reinforced instrumental behaviour.     

How children learn to draw realistic pictures

Children of ages 5 to 17 years were given a real scene to draw from a fixed viewpoint. Their drawings were classified according to the type of drawing system used, and given a score for the number of correct representations of occlusion by overlap. Six classes of drawing system used were identified, and it was found that with increasing age children were able to use an increasingly complex type of system. This development appeared to take place in discrete stages and covered the whole age range tested. Increase in the ability to use overlap as a pictorial device appeared to be continuous; below the age of 9 few children made use of overlap, and rapid learning took place between the ages of 10 and 12. It is argued that the use of these two pictorial devices requires different abilities; and that learning to use a particular type of drawing system does not depend directly on perception or on learning culturally determined stereotypes but in its demands on creative ability closely resembles the acquisition of language.     

Preschoolers use speakers’ preferences to learn words

In two studies, we investigated preschoolers’ ability to use others’ preferences to learn names for things. Two studies demonstrated that preschool children make smart use of others’ preferences. In the first study, preschool children only used information about others’ preferences when they were clearly linked to referential intentions. The second study established that children only used others’ preferences to learn words when the preference and intention information came from the same person. Both studies reveal impressive use of mental state information for word learning purposes.