Landau also pushes the role of syntax and its mapping to semantics in learning what some would consider “easy words”—the simplest spatial prepositions in English, in and on. Taking as a starting point that the syntactic distribution of a word is a reflex of its meaning, Landau shows that careful study of how children and adults linguistically encode a range of containment and support configurations reveals a special status for “core” configurations in each domain. She proposes that children come (...) to learning spatial terms with the expectation that configurations reflecting core concepts of containment and support will be expressed with the maximally simple, unmarked, linguistic expression. (shrink)

The performance of a connectionist learning system on a simple problem has been described by Hinton and is briefly reviewed here: a finite set is learned from a finite collection of finite sets, and the system generalizes correctly from partial information by finding simple "features" of the environment. For comparison, a very similar problem is formulated in the Gold paradigm of discrete learning functions. To get generalization similar to the connectionist system, a non-conservative learning strategy (...) is required. We define a simple, non-conservative strategy that generalizes like the connectionist system, finding simple "features" of the environment. By placing an arbitrary finite bound on the number and complexity of the features to be found, learning can be guaranteed relative to a probabilistic criterion of success. However, this approach to induction has essentially the same problems as many others that have failed. (shrink)

The Adaptive Toolbox framework specifies heuristics for choice and categorisation that search through cues in previously learned orders (Gigerenzer & Todd, 1999). We examined the learning of three cue parameters defining different orders: discrimination rate (DR) (the probability that a cue points to a unique choice), validity (the probability of correct choice given that a cue discriminates), and success (the probability of correct choice). Success orderings are identical to those by expected information gain (Klayman & Ha, 1987). In two (...) experiments, participants made choices in real-world environments with objective outcome criteria. Participant ratings indicated some appropriate parameter learning when the relevant cue parameter values were highly dispersed. Rated orders of cue validity and DR were less distinct than the objective orders—learning one parameter may be biased towards success by variation in the other. Success ratings capture the variation in validity and DR as well as participants' perception of these parameters. (shrink)

An agent who receives information in the form of an indicative conditional statement and who trusts her source will modify her credences to bring them in line with the conditional. I will argue that the agent, upon the acquisition of such information, should, in general, expand her prior credence function to an indeterminate posterior one; that is, to a set of credence functions. Two different ways the agent might interpret the conditional will be presented, and the properties of the resulting (...) indeterminate posteriors compared. The cause of the expansion from a single prior credence function to a set of credence functions forming the indeterminate posterior one will be explained. The expansion undermines the Bayesian dogma that the result of assimilating new information into a determinate prior credence functions is always a determinate posterior one. (shrink)

Although many studies have provided evidence that abstract knowledge can be acquired in artificial grammar learning, it remains unclear how abstract knowledge can be attained in sequence learning. To address this issue, we proposed a dual simple recurrent network model that includes a surface SRN encoding and predicting the surface properties of stimuli and an abstract SRN encoding and predicting the abstract properties of stimuli. The results of Simulations 1 and 2 showed that the DSRN model can (...) account for learning effects in the serial reaction time task under different conditions, and the manipulation of the contribution weight of each SRN accounted for the contribution of conscious and unconscious processes in inclusion and exclusion tests in previous studies. The results of human performance in Simulation 3 provided further evidence that people can implicitly learn both chunking and abstract knowledge in sequence learning, and the results of Simulation 3 confirmed that the DSRN model can account for how people implicitly acquire the two types of knowledge in sequence learning. These findings extend the learning ability of the SRN model and help understand how different types of knowledge can be acquired implicitly in sequence learning. (shrink)

Studies of conditioning in simple systems are best interpreted in terms of the formation of excitatory links. The mechanisms responsible for such conditioning contribute to the associative learning effects shown by more complex systems. If a dual-system approach is to be avoided, the best hope lies in developing standard associative theory to deal with phenomena said to show propositional learning.

The idea that “simplicity is a sign of truth”, and the related “Occam’s razor” principle, stating that, all other things being equal, simpler models should be preferred to more complex ones, have been long discussed in philosophy and science. We explore these ideas in the context of supervised machine learning, namely the branch of artificial intelligence that studies algorithms which balance simplicity and accuracy in order to effectively learn about the features of the underlying domain. Focusing on statistical (...) class='Hi'>learning theory, we show that situations exist for which a preference for simpler models provably slows down, instead of favoring, the supervised learning process. Our results shed new light on the relations between simplicity and truth approximation, which are briefly discussed in the context of both machine learning and the philosophy of science. (shrink)

Humans are incapable of acting as utility maximisers. In contrast, the evolutionary process had considerable time and computational power to select optimal heuristics from a set of alternatives. To view evolution as the optimising agent has revolutionised game theory. Gigerenzer and his co-workers can help us understand the circumstances under which evolution and learning achieve optimisation and Nash equilibria.

When making decisions, humans can observe many kinds of information about others' activities, but their effects on performance are not well understood. We investigated social learning strategies using a simple problem-solving task in which participants search a complex space, and each can view and imitate others' solutions. Results showed that participants combined multiple sources of information to guide learning, including payoffs of peers' solutions, popularity of solution elements among peers, similarity of peers' solutions to their own, and (...) relative payoffs from individual exploration. Furthermore, performance was positively associated with imitation rates at both the individual and group levels. When peers' payoffs were hidden, popularity and similarity biases reversed, participants searched more broadly and randomly, and both quality and equity of exploration suffered. We conclude that when peers' solutions can be effectively compared, imitation does not simply permit scrounging, but it can also facilitate propagation of good solutions for further cumulative exploration. (shrink)

The purpose of this study was to investigate experimentally the extent to which children’s novel word learning and story comprehension from eBooks depends on the relevance of interactive eBook features. A story was created in the lab to incorporate novel word-object pairs. The story was read to preschoolers (3-5 years old, N = 103) using one of the three books: noninteractive control, interactive-relevant, interactive-irrelevant. Novel word learning and story comprehension were assessed with posttests in which children picked target (...) objects from an array and sorted story events into sequences, respectively. Findings indicate that word learning and story comprehension were similar across all three books, suggesting that simple interactive features – whether relevant or irrelevant to the story – had little impact on learning in this controlled experiment. (shrink)

Dual process accounts of affective learning state that the learning of likes and dislikes reflects a learning mechanism that is distinct from the one reflected in expectancy learning, the learning of signal relationships, and has different empirical characteristics. Affective learning, for example, is said not to be affected by: (a) extinction training; (b) occasion setting; (c) cue competition; and (d) awareness of the CS-US contingencies. These predictions were tested in a series of experiments that (...) employed simple Pavlovian conditioning procedures. Neutral visual pictures of geometric shapes, or tactile conditional stimuli (CS) were paired with aversive electrotactile unconditional stimuli (US). Dependent measures were physiological (skin conductance, blink startle modulation) or verbal (US expectancy, on-line and off-line ratings of CS pleasantness). Different combinations of these dependent measures were employed across different experiments in an attempt to assess affective and expectancy learning simultaneously. Changes in CS pleasantness as indexed by ratings or blink startle modulation were readily observed. However, contrary to the predictions from dual-process accounts, results indicated that acquired CS unpleasantness is subject to extinction, occasion setting, cue competition, and not found in absence of CS-US contingency awareness. (shrink)

Simple idealized models seem to provide more understanding than opaque, complex, and hyper-realistic models. However, an increasing number of scientists are going in the opposite direction by utilizing opaque machine learning models to make predictions and draw inferences, suggesting that scientists are opting for models that have less potential for understanding. Are scientists trading understanding for some other epistemic or pragmatic good when they choose a machine learning model? Or are the assumptions behind why minimal models provide (...) understanding misguided? In this paper, using the case of deep neural networks, I argue that it is not the complexity or black box nature of a model that limits how much understanding the model provides. Instead, it is a lack of scientific and empirical evidence supporting the link that connects a model to the target phenomenon that primarily prohibits understanding. (shrink)

Using a simple learning agent, we show that learning self-control in the primrose path experiment does parallel learning cooperation in the prisoner's dilemma. But Rachlin's claim that “there is no essential difference between self-control and altruism” is too strong. Only iterated prisoner's dilemmas played against reciprocators are reduced to self-control problems. There is more to cooperation than self-control and even altruism in a strong sense.

This target article presents a theory of human cultural learning. Cultural learning is identified with those instances of social learning in which intersubjectivity or perspective-taking plays a vital role, both in the original learning process and in the resulting cognitive product. Cultural learning manifests itself in three forms during human ontogeny: imitative learning, instructed learning, and collaborative learning – in that order. Evidence is provided that this progression arises from the developmental ordering (...) of the underlying social-cognitive concepts and processes involved. Imitative learning relies on a concept of intentional agent and involves simple perspective-taking. Instructed learning relies on a concept of mental agent and involves alternating/coordinated perspective-taking (intersubjectivity). Collaborative learning relies on a concept of reflective agent and involves integrated perspective-taking (reflective intersubjectivity). A comparison of normal children, autistic children and wild and enculturated chimpanzees provides further evidence for these correlations between social cognition and cultural learning. Cultural learning is a uniquely human form of social learning that allows for a fidelity of transmission of behaviors and information among conspecifics not possible in other forms of social learning, thereby providing the psychological basis for cultural evolution. (shrink)