Theory and operational definitions in computational memory models: A response to Glenberg and Robertson

Glenberg and Robertson (2000) conducted three experiments to investigate th e adequacy of high-dimensional memory models in accounting for the meaningf ul interpretation of sentences. They conclude that high-dimensional memory models (such as LSA and HAL) are inadequate as theories of meaning. As an a lternative account, they offer an embodied approach to cognition-the indexi cal hypothesis. In this article, I detail how Glenberg and Robertson have f ailed to extend to the models they criticize the same considerations that a re extended to the humans in their experiments. The model is evaluated with out providing appropriate experience for the tests that are conducted. More over, testing a representational model (e.g., LSA) for processing considera tions is inappropriate and ultimately contributes to what are serious flaws in their interpretation of their experimental results. (C) 2000 Academic P ress.