Model-dependent realism

In philosophy, model-dependent realism asserts that there is no "objective" or "unique" external reality, but reality consists at least in part of networks of world pictures that connect observations with their explanations or models.

A world picture consists of the combination of a set of observations accompanied by a conceptual model and by rules connecting the model concepts to the observations. Different world pictures that describe particular data equally well all have equal claims to be valid. There is no requirement that a world picture be unique, or even that the data selected include all available observations. The universe of all observations possibly may be covered by a network of overlapping world pictures and, where overlap occurs; multiple, equally valid, world pictures exist.

Outline
Recently the connection between models and observations has been explored by physicists Stephen W. Hawking and Leonard Mlodinow in their book: The Grand Design, where they propose the notion of model-dependent realism, a view of reality based upon world pictures. Hawking/Mlodinow introduce a world picture as follows: This quotation defines a "world picture". This concept of "world picture" is extended to a wider class of mental constructs in the metaphysical position of model-dependent realism as proposed below: Hawking/Mlodinow point out:


 * that either an earth-centered (Ptolemaic) or a sun-centered (Copernican) picture of reality can be made consistent with the motion of celestial bodies;
 * that goldfish physicists living in a curved bowl, though observing curved paths of motion of bodies that we observe as linear, could still formulate predictive laws governing motion as they see it;
 * that we cannot know whether we live in a simulated world, a virtual reality, one that the simulators rendered self-consistent.

Each of those world pictures is not only data-dependent, but is explanation-dependent.

In comparing different world pictures, If two different models agree with the observations, to consider one more true than the other, to say that one gives a truer picture of reality than the other, involves considerations outside the concept of model-independent reality, to consider possibly subjective matters such as which may be more convenient to employ in a given situation, more elegant, more intriguing, or otherwise more appealing.

Some find the ambiguity introduced by alternative equivalent world pictures to be a defect of the concept of model-dependent realism. That position, however, is based upon outside criteria, and an example of such criteria is provided shortly.

It should be emphasized that there is no restriction that a model use only observable or measurable constructs. The alternatives: are addressed by Hawking/Mlodinow in their definition of a model as follows:

Therefore, the definition of a model adopts any unobservable constructs as aspects of the model. With this interpretation of a model, the alternatives posed by Cao above are confronted by the metaphysical position of model-dependent realism first by rejection of any posit of "objective" reality beyond world pictures or networks of world pictures that organize experience, and second, like the world pictures themselves, model-dependent realism accepts any unobservable entities in the models used by these world pictures as part of the reality included in model-dependent realism.

Background
The "reality" of science, even when restricted to the interpretation of observations and measurements, has been much discussed. Pierre Duhem (1861-1916) held that a physical model is no more than an aid to memory, summarizing and classifying facts by providing a symbolic representation of them and, moreover, the facts of physical theory are to be distinguished from common sense and metaphysics. His views were further developed by W. V. O. Quine (1908-2000), who suggested "“our statements about the external world face the tribunal of sense experience not individually, but only as a corporate body”. It is impossible to test a scientific hypothesis in isolation, but only as part of a system. These two authors were much concerned with how a theory was coupled to concrete observation and measurement, and how it morphed with admission of new data.

The evolution of science forms part of this discussion. For example, Thomas Kuhn connected changes in scientists' views of reality to "revolutions" in science and changes in "paradigms". As an example, Kuhn suggested that the Copernican "revolution" replaced the views of Ptolemy not because of empirical failures, but because of a new "paradigm" that exerted control over what scientists felt to be the more fruitful way to pursue their goals. Such historical analysis goes beyond the concept of a world picture itself to critique competing world pictures, and to assess how changes in ways to evaluate a world picture affect its influence over the development of knowledge.

The matter is made more complicated by attempts to extend observations of scientific practice to wider realms, including religious systems, in an attempt to compare them. A key author in this arena was Barbour who proposed an approach called critical realism. The word "critical" refers to reflection and analysis. This broad extension lies outside the realm of model-dependent realism itself, and falls into a much vaguer and more tendentious arena.

Model assessment
Many world pictures or theories may be proposed, and the issue of comparing, ranking, and generally critiquing them arises.

For example, quantum mechanics, which is a world picture or theory describing (among other matters) atomic interactions, despite its experimental success, is commonly called incomplete as it is "not accompanied by an interpretation that is widely convincing." Steven Pinker discusses this question using several quotations, including one from Murray Gell-Mann that describes quantum theory as: "that mysterious, confusing discipline which none of us really understands but which we know how to use." These reservations about quantum mechanics appear to seek something more than a world picture, what might be called physical intuition, or visualization.

Hawking/Mlodinov do not address the intuitive qualities of a world picture or theory, or scientists' personal opinions of them, but they do raise the question of what constitutes a good theory. They suggest a good theory has these characteristics:(p. 51 )
 * 1) It is elegant
 * 2) Contains few arbitrary or adjustable elements
 * 3) Agrees with and explains all existing observations
 * 4) Makes detailed predictions about future observations that can disprove or falsify the model if they are not borne out.

The features of a "good" theory have been debated for centuries. The last criterion on this list is related to the criterion proposed by Popper: Five somewhat similar criteria were proposed by Kuhn as what he called "the shared basis for theory choice", a list selected "not because they are exhaustive, but because they are individually important and collectively sufficiently varied to indicate what is at stake."

These desiderata of a "good theory" allow a critique of different theories. These "principles of comparison" may have a justification beyond mere general acceptance.

Unfortunately, even the most successful world picture of modern science, the Standard Model of particle physics, satisfies only the last criterion. As said by Hawking/Mlodinov (p. 52 ): The Standard Model fails the third criterion in not encompassing gravitation. Hawking/Mlodinov (p. 58 ) deal with the failure of a theory to encompass all observations using the notion of a network of overlapping theories, each describing some observations and agreeing with one another where the theories overlap. To quote:(p. 58 ):

Presumably, each theory or world picture included in a network provides concepts for a "model-dependent reality", though that reality is restricted to the domain of data to which it applies. Where these model-dependent realities overlap, multiple interpretations of reality are available of equal value.

Data collection
The definition of a model given by Hawking/Mlodinow is pretty straightforward if one has in mind a particular set of data to explain. Either the model explains the data or it doesn't, and if two models explain the data differently, any claim for the concepts employed by either as more true of "reality" must be based upon criteria lying outside the reach of model-dependent reality, such as the desiderata for a "good" theory listed earlier.

The matter is less clear when one considers the selection of just what "data" must be explained. Our senses are limited, and we accept that we cannot see and hear everything that comprises reality. So we supplement the senses, for example, by using a telescope or a microscope. Historically the issue arose as to whether such instruments deceived us, and gradually they have been accepted as extensions of our natural capacities.

The gathering of "data" supplementing our senses has gone far beyond the primitive telescope to its modern version (for example, the Hubble telescope) and the microscope to its modern version (for example, the scanning tunneling microscope). Today experiments may require expensive apparatus not available to all, involving observations not even interpretable by many. Examples are the colliders of high-energy physics, and the sophisticated electronic image acquisition of modern astronomy, guided by elaborate computer processing and filtering. One might reasonably ask how well the acquisition of "data" is separated from the "theory" that explains how the acquisition process works, and that often suggests where to look for new "data". The process by which data is allowed into the theory influences what is incorporated into "reality".

The gathering of data is complicated by the limited access to these data-acquisition instruments, both in a required training that could be seen as indoctrination (not necessarily deliberate, but de facto), and in limitations upon who, and what investigations, are worthy to use the instruments, as determined by various funding agencies and corporate laboratories. Although censorship is not the motivation directing government and corporate support, a preoccupation with popular and/or commercially attractive projects draws resources and talent away from less conspicuous goals potentially of more significance to a comprehensive "reality". In effect, the expense and expertise of modern research result in blinkers.

The analysis as well as the gathering of data is becoming more complicated as our very notion of thinking, even of mathematical proof, is modified by technology, for example, by computers. Theoretical predictions are made by computer simulations that perform calculations beyond human capacity. The concepts entering a model-based reality may be only implicit in a computer programmable code, in open-ended algorithms, and may not be concepts the human mind is aware of directly.

To a limited degree, the shaping of "reality" based upon modeling of selected data is a public enterprise, with all the foibles that implies. The public does not engage reality at a specialized deeply technical level, but at a metaphoric level:

An explicitly metaphoric public participation is "eco-consciousness". Metaphorical involvement also is evident in arenas such as gene research and genetically altered organisms, and investigations of stem cells, where the public is actively engaged. Another example is archaeology and the limitations exerted upon examination of burial sites. In some cases public participation leads to simple clamor, as in the case of global warming. This broad public engagement, frequently informed by vested interests and oversimplifications, facilitates manipulation by groups with their own objectives, similar to the censorship found in the times of Vesalius and Galileo although lacking some of that institutional authority.

Although the above examples suggest an indictment of metaphor as a foible of public participation in shaping reality, public engagement in some form is necessary and desirable, and ultimately a goal of the entire enterprise.