William Baggett

The Time Course of Generating Causal Antecedent and Causal Consequence Inferences.

The present study tested whether causal antecedent and causal consequence inferences are generated on‐line during comprehension and also determined the time course of their activation. The study manipulated inference category, the rate of word presentation in a rapid serial visual presentation (RSVP) format, and the delay between the last word in a sentence and the test word (i.e., SOA interval). Lexical decision latencies were collected on test strings (i.e., nonwords, inference words, or unrelated words) which were presented after each sentence in the passages. The results indicated that there was a threshold of 400 ms after stimulus presentation (RSVP and SOA) before causal antecedents were generated on‐line, whereas causal consequences were not generated on‐line. These results support a bridging model of inference generation which assumes that causal antecedent inferences are needed to bridge an explicit text event with prior passage content, whereas expectation inferences are not normally generated o...

Question‐driven Explanatory Reasoning

The primary claim in this paper is that questions are one of the fundamental cognitive components that guide human reasoning. That is, threads of coherent reasoning are built around the questions that humans ask and their answers to these questions. Explanatory reasoning is elicited by particular classes of questions (such as why, how, and what-if) that invite the construction of causal chains, goal-plan-action hierarchies, and logical justifications. This paper identifies the psychological mechanisms that underlie human question asking and question answering, along with some empirical findings that support these mechanisms. We also discuss some ways that educational software can be designed to facilitate question-driven explanatory reasoning.

Answering when questions about future events in the context of a calendar system

Three experiments tested a model of question answering called WHEN, which explains the answer descriptions that are generated when adults answer when questions (Golding, Magliano, & Hemphill, 1992). College students answered questions about future events in the context of a 12‐month calendar year. The WHEN model specifies how the time of the future event is expressed as a function of the temporal interval between the present point in time and the time of a future event (1–90 days away). The answers included generative descriptions (e.g., “next week on Wednesday”) and specific dates (e.g., “August 13”). The answers systematically varied as a function of temporal interval in a fashion that supported most of the production rules of the WHEN model.

Macro-adaptation in Conversational Intelligent Tutoring Matters

We present in this paper the findings of a study on the role of macro-adaptation in conversational intelligent tutoring. Macro-adaptivity refers to a systems capability to select appropriate instructional tasks for the learner to work on. Micro-adaptivity refers to a systems capability to adapt its scaffolding while the learner is working on a particular task. We compared an intelligent tutoring system that offers both macro- and micro-adaptivity fully-adaptive with an intelligent tutoring system that offers only micro-adaptivity. Experimental data analysis revealed that learning gains were significantly higher for students randomly assigned to the fully-adaptive intelligent tutor condition compared to the micro-adaptive-only condition.

Design and testing of a general-purpose neurocomputer

Abstract Here we describe the logical design and testing of a general-purpose neurocomputer, AMNIAC. It may be thought of as a programmable neural network that can simulate arbitrary SIMD and MIMD machines of practical interest (modeled as cellular automata, neural networks, or arbitrary automata networks). AMNIAC is purely bitwise (amnesic), i.e., requires no local memory or registers, other than short memory just long enough for a clock cycle. We discuss software serial and massively parallel simulations of AMNIAC as (a) tests of the logical design; and (b) benchmarks for evaluation of the trade-off between its universality and memory advantage versus overhead cost of mapping and speed. Theoretical applications of the design are given. A 3D SIMD version of AMNIAC which stabilizes if and only if its input network stabilizes (on the same input data) establishes the unsolvability of the stability problem for networks of finite bandwidth and also its weak solvability by an extension to a larger activation set. We also discuss the feasibility and trade-offs of a physical implementation.