The accusation that I work under the assumption that spatial models are either the most important or the only type of situation model people construct while reading a narrative text couldn't be further from the truth. What I wrote about empirical research on inferences in the spatial domain is that it makes a particular pedagogical point. It happens to be extremely easy to distinguish between a representation of the linguistic form of a text about a spatial layout and a representation of the situation it describes. Hence, both a priori arguments and empirical evidence for the use of mental models are easy to find in the spatial domain. Mental-models theory is supposed to be a theory of how ALL texts are understood. Similarities between the structure of a text and the structure of the situation it describes do not always exist. But even when there are similarities between the structure of a text and the structure in the world that it describes, there are crucial differences between mental models and representations of text structure.
2. For a particular reader of a particular text (perhaps with a particular goal), spatial inferences may be more or less important. The subject matter of the text is, obviously, a crucial determinant of the role that spatial inferences play in understanding it. I have long endorsed (see e.g. Garnham, 1985: 177) Miller and Johnson-Laird's (1976) suggestion that, in text comprehension, it is necessary to establish the logical, temporal, causal, intentional and moral connections between the events, states and processes described in different parts of the text, as well as the spatial ones. In the mental model of a particular text, relations of several of these kinds are likely to be encoded. It would be a rare text (and a rare reader, with a rare goal) that required a mental model that encoded only spatial relations -- one that warranted the appellation "spatial model." Furthermore --and here I agree with Zwaan and Graesser (1993: 3) -- there are reasons for thinking that spatial models might be comparatively difficult to construct. There are, however, counterarguments. The spatial domain is a particularly favored source of metaphors (Lakoff & Johnson, 1980), suggesting that people find processing spatial information, and in particular spatial relations, relatively easy.
3. I am also surprised by Haberlandt's claim that "it is not clear how mental models would capture... causal links" (1993: 3). Mental models contain information about particular events, and they may directly encode relations between them. Of course, if one believes that mental models are "life-like," and nothing else, there may be a problem with causal relations. As Hume pointed out long ago, we do not experience causation directly (though cf. Michotte's work [e.g., 1954]), but only the conjunction of events in the world. If a mental model simply represented a sequence of events (a ball hitting a window; the window breaking), there would be no representation of the causal relation in the model. Even if causation is explicitly signaled ("the window broke because the ball hit it"), it is not clear how it would be represented in a "life-like" model.
4. There are two possibilities, and I don't know which is right. Indeed, it is possible that each is realized under different circumstances. The first is that causality is represented by links between the representations of events. We know that mental models have to have various adjuncts to their representations of situations in the world. For example, they need to be supplemented with information about the people to whom they are addressed -- the particular person one is talking or writing to, or the general audience of a public lecture or a written text (see Johnson-Laird & Garnham, 1980). One problem with this kind of adjunct in the case of causation is apparent from the discussion of Noordman and Vonk's (1992) data (see e.g. Garnham & Oakhill, 1992). Simply encoding that events, states and processes are somehow or other causally linked does not result in an adequate representation of the causal connection between them.
5. The second possibility is that causal relations are not encoded directly into mental models. Rather, when a model is constructed, the reader or listener should establish that the encoded events are related, causally or otherwise, in a way that makes sense in terms of that person's knowledge about how the world (or the world that the text is about) works. Thus, as in my omission theory of case filling inferences (Garnham, 1982), information about causal relations can be left out of a mental model if it can be filled in later using knowledge stored in long-term memory. However, although this view is plausible for simple causal relations between two events, its extension to complex causal chains would require a different type of checking process. Information about causal chains presumably cannot be stored as a set of templates but must be held in the form of a "grammar" of causation. So deciding whether a complex causal chain is consistent with background knowledge would require a process akin to parsing.
6. I would like to close with a reminder that the mental-models theory is supposed to be a theory of how ALL texts are understood, not just texts about space and, indeed, not just texts about concrete situations. Sometimes there are similarities between the structure of a text and the structure of the situation it describes, for example, between a linear sequence of causally related events and a text that describes them in the order in which they occurred. Sometimes there are no such similarities, for example, in texts about spatial layouts. But even when there are similarities between the structure of a text and the structure in the world that it describes, there are crucial differences between mental models and representations of text structure. Texts are made up of words that are syntactically and semantically related to one another. A representation of their structure would include elements standing for words and other linguistic entities and elements standing for the syntactic and semantic relations. A mental model is a representation of a situation and contains elements that represent people, things, events, states, processes, times and places, and real world relations between them.
Garnham, A. (1982). Testing psychological theories about inference making. Memory and Cognition, 10, 341-349.
Garnham, A. (1985). Psycholinguistics: Central topics. London: Methuen.
Garnham, A. (1992). Minimalism versus constructionism: A false dichotomy in theories of inference in reading. PSYCOLOQUY 3(63) reading-inference-1.1
Garnham, A. & Oakhill, J.V. (1992). Discourse representation and text processing from a "mental models" perspective. Language and Cognitive Processes, 7, 193-204.
Glenberg, A.M. & Mathew, S. (1992). When minimalism is not enough: Mental models in reading comprehension. PSYCOLOQUY 3(64) reading-inference-2.1.
Haberlandt, K. (1993). Understanding mental models and inferences. PSYCOLOQUY 4(5) reading-inference.6
Johnson-Laird, P.N. & Garnham, A. (1980) Descriptions and discourse models. Linguistics and Philosophy, 7, 391-
Lakoff, G. & Johnson, M. (1980). Metaphors we live by. Chicago: Chicago University Press.
Michotte, A. E. (1954) La Perception de la Causalite. Studia Psychologica: Publications Universitaires de Louvain.
Miller, G.A. & Johnson-Laird, P.N. (1976). Language and perception. Cambridge: Cambridge University Press.
Noordman, L.G.M. & Vonk, W. (1992). Reader's knowledge and the control of inferences in reasoning. Language and Cognition, 7, 373-391.
Zwaan, R.A. & Graesser, A.C. (1993). Reading goals and situation models. PSYCOLOQUY 4(3) reading-inference.4.
AUTHOR NOTE: My work on mental models has been supported by ESRC grant C 0023 2439 "Mental models and the interpretation of anaphora. Thanks to Jane Oakhill for comments on an earlier draft.