The Development of Geocentric versus Heliocentric Models of the Solar System: The Predictive versus Explanatory Powers of Models

        The predictive power of a model does not always mean that it is accurately represents the true nature of what it is trying to explain. The geocentric model of the solar system is an example of such a case. Geocentric models dominated astrology and astronomy for almost 2000 years before scientists realized that these models, despite their predictive power, did not correctly describe the nature of the stars, the planets, and their movement. And even when the proper model, the heliocentric model, was discovered, it was not fully accepted until it produced better predictions. This case calls into question whether predictive capability alone should allow a certain model to dominate the discourse of a certain research program. It also suggests that scientists in general should be more open to non-conventional models in their discipline, even if these models do not immediately provide better predictions than the conventional models.

More after the jump…

        It is not surprising that geocentric models of the solar system were dominant for several millennia. First, based on simple observation, it appears as though the sun, stars, and planets rotate around the earth. Second, it is difficult to perceive that the earth actually moves itself. Hence the earth seems stationary while everything in the sky moves around the earth. Furthermore, not only did simple observations of the sky re-enforce the idea that the earth was the center of the solar system, religion did as well. Both Catholic and Islamic religious scholars both agreed that we lived in a geocentric solar system. 
        

        Given that both basic observation and religious doctrine re-enforced the geocentric model of the solar system, it was not until the middle of the 16th century that scientists (namely Copernicus) began considering heliocentric models. The key to the discovery and acceptance of this accurate model of the solar system relied on explaining one strange phenomenon that geocentric models could not accurately predict and explain: the occasional retrograde motion of planets in the night sky. With retrograde motion, a planet briefly moves from west to east in the night sky, instead of the usual east to west motion (see here for a more detailed explanation as to why this occurs).
        

        While most geocentric models of the solar system, including those dating back to both Aristotle (350 B.C.) and Ptolemy’s famous model in Almagest (140 A.D.), try to account for this retrograde motion (see here for their explanations), they still could not accurately predict the timing of this motion. Furthermore, even earlier heliocentric models could not better predict the timing of this odd movement of than the geocentric version (as well as the general movement of the planets and stars), even though the heliocentric models were more accurate representations of the solar system!
       

        It was not until the heliocentric models of the solar system were more fully developed that they produced better predictions. For example, one of the flaws of Copernicus’s early model is that he assumed the planets orbited the sun in a circular model. Kepler (17th century) further refined this model, and instead assumed the planets orbited the sun in as elliptical motion of planets around the sun. This provided more accurate predictions of the motion of planets, especially Mars. 
        

        While Kepler’s work was also not fully accepted initially (in some part due to the limits of mathematics at the time limiting his ability to make perfect predictions), it eventually spread throughout Europe. Newton, for example, used calculus to derive Kepler’s laws of planetary motion from Newton’s own laws of motion and gravitation.
        

        The development of geocentric and heliocentric models of the solar system demonstrate how the predictive power of a model, as well as the inability of being open minded about non-conventional models, can mislead discourse for a long period of time. This is not to say the predictive power of a model should not be taken into account. Predictive capability is an important component of comparing models, but it is not the only component.  Given that mathematics is limited in its ability to only demonstrate correlation, not causation, it is essential to always be skeptical of statistical results and open to the possibility that another explanation may exist.