The authors outline a cognitive and computational account of causal learning in children. They propose that children use specialized cognitive systems that allow them to recover an accurate "causal map" of the world: an abstract, coherent, learned representation of the causal relations among events. This kind of knowledge can be perspicuously understood in terms of the formalism of directed graphical causal models, or Bayes nets. Children's causal learning and inference may involve computations similar to those for learning causal Bayes nets and for predicting with them. Experimental results suggest that 2-to 4-year-old children construct new causal maps and that their learning is consistent with the Bayes net formalism.The input that reaches children from the world is concrete, particular, and limited. Yet, adults have abstract, coherent, and largely veridical representations of the world. The great epistemological question of cognitive development is how human beings get from one place to the other: How do children learn so much about the world so quickly and effortlessly? In the past 30 years, cognitive developmentalists have demonstrated that there are systematic changes in children's knowledge of the world. However, psychologists know much less about the representations that underlie that knowledge and the learning mechanisms that underlie changes in that knowledge.In this article, we outline one type of representation and several related types of learning mechanisms that may play a particularly important role in cognitive development. The representations are of the causal structure of the world, and the learning mechanisms involve a particularly powerful type of causal inference. Causal knowledge is important for several reasons. Knowing about causal structure permits us to make wide-ranging predictions about future events. Even more important, knowing about causal structure allows us to intervene in the world to bring about new eventsoften events that are far removed from the interventions themselves.

Algorithms play a key role in the functioning of autonomous systems, and so concerns have periodically been raised about the possibility of algorithmic bias. However, debates in this area have been hampered by different meanings and uses of the term, "bias." It is sometimes used as a purely descriptive term, sometimes as a pejorative term, and such variations can promote confusion and hamper discussions about when and how to respond to algorithmic bias. In this paper, we first provide a taxonomy of different types and sources of algorithmic bias, with a focus on their different impacts on the proper functioning of autonomous systems. We then use this taxonomy to distinguish between algorithmic biases that are neutral or unobjectionable, and those that are problematic in some way and require a response. In some cases, there are technological or algorithmic adjustments that developers can use to compensate for problematic bias. In other cases, however, responses require adjustments by the agent, whether human or autonomous system, who uses the results of the algorithm. There is no "one size fits all" solution to algorithmic bias.

The Rescorla-Wagner model has been a leading theory of animal causal induction for nearly 30 years, and human causal induction for the past 15 years. Recent theories (especially Psychol. Rev. 104 (1997) 367) have provided alternative explanations of how people draw causal conclusions from covariational data. However, theoretical attempts to compare the Rescorla-Wagner model with more recent models have been hampered by the fact that the Rescorla-Wagner model is an algorithmic theory, while the more recent theories are all computational. This paper provides a detailed derivation of the long-run behavior of the RescorlaWagner model under a wide range of parameters and experimental setups, so that the model can be compared with computational theories. It also shows that the model agrees with competing theories on a wider range of cases than had previously been thought. The paper concludes by showing how recently suggested modifications of the Rescorla-Wagner model impact the long-run behavior of the model. r

No abstract