摘要：In the theoretical framework developed by Johnson-Laird (1983; 1993) about mental models and their role in human reasoning, induction is defined as a thought process starting with a set of observations, whose goal is to frame a hypothesis that reaches a better description or understanding of this information in relation to a background of general knowledge; any hypothesis can be evaluated and as a result, maintained, modified or abandoned. A distinctive feature of induction is that its conclusions increase knowledge in a plausible way. The J ohnson-Laird theory describes how such a process generates hypotheses through procedures of construction, manipulation and revision of mental models; it does not, however, give an explicit and formal (i.e., useful in the design of a computational model) description about how conclusions can go beyond the specific events considered to enrich the initial information
