"Induction seeks facts to test a hypothesis; abduction seeks a hypothesis to account for facts." – Charles Sanders Peirce

In a previous post touching on the philosophical school of Pragmatism, Charles Sanders Peirce was identified as one it’s founders. One of the core concepts he developed through pragmatism was abductive reasoning. Here a few related ideas will be noted to enhance the more general ideas of Pragmatism.

In essence, abductive reasoning is the process of observing a fact and then determining the cause of that fact. For example, if it is observed upon waking that the grass is wet, it is reasonable to suppose that it rained the previous night. An informal deductive logic argument be: a) when it rains, things become wet b) it is raining on the grass c) the grass is wet. An inductive logic argument might be formed: when it rains on the grass, the grass often becomes wet; it will rain tonight; the grass should become wet.

The abductive logic argument is not intended to be necessarily correct. There are a number of ways the grass could have become wet, but, barring some additional knowledge (e.g. the sprinkler system was on), the simplest, most economical explanation is that it rained the previous night. This simple choice of hypothesis is at the core of Pragmatism. No matter which form of reasoning is employed, some narrowing down of criteria for the validity of an explanation is always required and Peirce advocates for starting with the most economical, easily tested choice, similar in spirit to Occam’s razor What remains is for the hypothesis to be adequately tested through the scientific method and a consensus of competent inquirers should agree on the outcome.

Developed by Peirce in the second half of the 19th century and early 20th, the adoption of this economical and flexible logic represented a great departure from classical logical arguments. The history of Philosophy contains countless demonstrations starting from the concepts of reason, e.g. Descartes proof of the the existence of God: a) I have an idea of God, a perfect being; b) There must be as much reality or perfection in the cause of any thing as in the effect; c) The idea of God represents something so perfect that I could not have been the cause of this idea.d) Therefore, God must exist as the only possible cause of the perfection found in my idea of Him.)

Alternatively, inductive arguments observing that a) Starving children are dying slow agonizing deaths b) God is all-loving c) An All-Loving God would not create a world with children dying through starvation d) An All-Loving God does not exist. An abductive argument would require some fact as a starting point for hypothesis formation which is in part why Peirce did not much take up the argument of God, though an abductive reasoner would likely advocate for economical, testable hypotheses when presented with claims for miracles. The divinity question itself presented a bit of a split in the Pragmatist school, with William James arguing for personal interpretations of divinity and Peirce recommending to avoid the topic within the bounds of pragmatism.

The applications for abductive reasoning having been many, including Artificial Intelligence (e.g. fault diagnosis), Medicine (illness diagnosis), Linguistics, Philosophy of Science, Anthropology and many more.

.In the realm of Computer Science, there is a role for heuristics, guessing and abductive reasoning. In the following video clip, you can see an MIT instructor teaching students the art of Guessing when it comes to solving the problem of developing a procedure to calculate the shortest path between two points separated by any number of nodes. By trying every possible solution once, it can be determined that there is at least one path, but it’s far from certain to be the shortest. In creating hypothesis to test the various paths, there are endless choices. One could add up all the distances between the nodes and average them or starting examining the distance between the 2nd and 3rd nodes, and on and on.

The instructor encourages the class to guess where to start and one student suggests looking at the distance of the closest nodes from the starting point. The guess is good, but as it turns out there is a slightly better guess – starting from the endpoint. In any event, this is an example how economical hypothesizing provides a pragmatic entry point for testing.


In the references provided below, much more detailed information about the history, testing and application of abductive reasoning can be found, along with extensive bibliographies for additional reading.