In this third part of my review of The Grand Design (Stephen Hawking & Leonard Mlodinow), I’d like to take a closer look at their theory of “model-dependent realism”(42).
This is one of Hawking’s and Mlodinow’s main contributions in The Grand Design. According to this theory:
a physical theory or world-picture is a model (generally of a mathematical nature) and a set of rules that connect the elements of the model to observations. This provides a framework with which to interpret modern science. (43)
In their view “to model-dependent realism it is pointless to ask whether a model is real, only whether it agrees with observation” (46); moreover, if two different models agree with observations, we cannot know which one is “more real than another” (46). This is only wordplay – saying that a “model is real” means precisely that it agrees with observation.
What else could it mean? How can a model “agree with observation” of nature and not be “real” i.e. not correspond to the reality it models? If a model makes testable predictions which are validated, the very fact of validation means something about the model is correct, i.e. corresponds to reality and, therefore, provides real information about nature. Clearly, Hawking and Mlodinow are trying to establish a distinction between ‘agreeing with observation’ and ‘being real’ – but what purpose can such a distinction serve in the pursuit of science? Imagine it being applied in virology: our model correctly predicts the nature and behavior of a deadly virus – but that’s no reason to make a vaccine, since the agreement with the model doesn’t tell us anything real about the virus. Who would accept such reasoning, let alone act on it?
The ultimate, and devastating consequence of accepting this line of thought is that it makes science impossible. Science is no longer a quest for knowledge about the world or nature; it is the quest for knowledge about our theories or models of the world – which is a very different thing. According to model-dependent realism, there is no such thing as scientific knowledge of nature but only knowledge of our own models. And even that is undermined by degeneration into an infinite regress, for when we check a model against our observations, we must also have a model of what constitutes ‘an observation’ and that model requires further observations which in turn must be checked against our model and so on. Furthermore, we cannot even know our own models, because to make a model we have to have a model of models, (and observe whether our model of models agrees with the models we check) and then a model of the model of models and so on ad infinitum.
This catches Hawking and Mlodinow in a logical tangle from which there is no escape: we cannot know nature (as they admit), but neither can we really know what a ‘model’ or an ‘observation’ is. The clear upshot is that science as the quest for knowledge about nature is impossible.
Hence, Hawking’s view covertly carries within it a profound and corrosive skepticism about the possibility of real knowledge about nature. All we can know are our models – and ultimately, as we have seen above, not even those. Moreover, if all knowledge is model-dependent, can we know anything about anything since all we can really know is whether or not our observations agree with our model? It is, after all, “pointless to ask whether a model is real” (46) i.e. whether a model gives knowledge about reality. This skepticism is precisely why Hawking and Mlodinow can undermine the whole concept of progress in science by claiming that the Copernican model is merely more convenient and not more correct than the Ptolemaic model of the solar system. There is no progress because there is no true model or knowledge about nature – only more or less convenient models for whatever our purpose happens to be. That, of course, reduces ‘truth’ to whatever we want it to be.
Next time: Whence “model-dependent realism”?