Reasonings About Philodemus’ Treatise On Methods of Inference

On Methods of Inference is the only extant work on logic that I know of that exists within our tradition. In it, Philodemus was arguing against other philosophers, mostly against Stoics, who relied heavily on syllogisms and logical juggling in order to ascertain truths.

There are many difficulties with this work, chief among them the fact that Epicureanism is a materialism and by necessity highly empirical, based heavily on evidence presented before the tribunal of the senses and things that can be grasped with our faculties as natural beings, whereas other philosophies are happy to discard the facticity of matter and evidence and degenerate into speculation and irrelevant abstraction.

On the Canon

There are several key features to our approach to logic, but prior to addressing them we must address why reason and logic are not within the canon. Epicureanism is a materialist and realist tradition of philosophy according to which reality exists and can be grasped with our faculties.

Because truth can be grasped, we are dogmatists: we believe that new truths must be in symphony with previously established truths, and that there must be criteria by which we judge things to be real.

These criteria are the canon (or measuring stick), our connection to reality. The canon includes three sets of faculties: the five senses, the pleasure/aversion principle, and the anticipations.

So far, so good. Now, some have interpreted anticipations as pre-conceptions, and given the impression that anticipations are ideas, concepts that are formed by the rational mind, but we hold this to be an error. Anticipations are innate, pre-rational. The three sets of faculties in the canon give raw data to reason and it’s the job of reason to make sense of it, to calculate with this data, but the data itself is irrational and whatever error we make with regards to this data, is an error of calculation, of reason, of logic. This is why logic/reason is not within the canon.

On the issue of how anticipations are innate and pre-rational, our friend Cassius, while citing Thomas Jefferson in his thoughts on anticipations, the canon, and reason, makes the case that humans, through natural selection, evolved as social beings with certain faculties. He also says:

The problem with the conventional Bailey view is that they say this “faculty” does not exist and they say that what Epicurus was talking about was the conceptual reasoning process. They say that anticipations are :

1) I see a chair

2) I see another chair

3) I form a concept in my mind of what I THINK a chair ought to look like

4) BINGO — that is an “anticipation” and the next time I see a chair, I recognize it because the “anticipation of a chair” is stored in my mind

Dewitt says that is ALL WRONG — because our mind, using reasoning in that case, have formed a CONCEPT of a chair.

See also this article on Norman DeWitt, who said: “It was Epicurus’ determination to dethrone reason and set up nature as the norm.”

And so Epicureanism is fundamentally different from much reason-based philosophy out there. We’re very unlike the Aristotelians, the objectivists, and other cults of reason out there. We always keep our feet on the ground and do not consider irrelevant speculation without evidence to be true philosophy. Unlike other schools of philosophy, we look to Venus more than Athena.

And so the first problem that must be addressed when we approach On Methods of Inference is that we reject the premise that logic is a criterion for truth, as it was to the philosophers that Philodemus was arguing against. Cassius states: “You need to see the basic argument without letting them fool you into thinking they know more than you do”.

This is the kind of difficulty one encounters when one argues with non-Epicurean philosophers, or when one studies the arguments between various schools that utilized very different points of reference.

The Basic Arguments

Our friend Brian made available the following chart to help us begin to grasp the treatise. It includes some basic definitions, first among them the idea of signs, which are symptoms or indications that something is there and is real.

From a sign, we get what we see (which falls within the canon) and what we can infer (which does not and may be subject to error). As I see it, there key controversies in the text have to do with how we can know with any certainty that our inferences lead to firm truth, since they constitute indirect knowledge that is not immediately rooted in our experience or in evidence. Inferences can be of two types:

  • general inferences draw generic information from a single instance of concrete fact
  • particular inferences draw very specific conclusions about a concrete fact

Our friend Cassius states: “… the Epicurean method of establishing truth is reasoning by analogy, by extention, and by comparison to things that we know already because they are clear … and … the method of Epicurus is sort of measurement-based“.

One of the main objections raised by other schools had to do with how Epicureans frequently draw generic non-evident conclusions from concrete, observed phenomena. These discussions centered on how there are many unique cases and exceptions to things that are generally perceived.

To justify this tendency to make generalizations, Epicureans cited inconceivability, which we can simplify by saying they invoked common sense. According to our friend Cassius, “inconceivable probably refers to something that CONFLICTS WITH A KNOWN FACT“. This is vindicated by Philodemus himself on the scroll:

… this has been observed to be a property of that in all cases that we have come upon, and because we have observed many varied living creatures of the same genus who have differences in all other respects from each other, but who all share in certain common qualities (e.g. mortality). – On Methods of Inference XXXV

For instance, it’s inconceivable that a man may be immortal because all men who existed in the past have died and there are no known and empirically verified cases to the contrary, and so it’s fair to expect that all men living today will also die.

We may call this the argument of no known exceptions: since all men are known to die, and we have no reason to suspect that men outside of our direct experience are immortal, then we can conclude that all men are mortal. The words used by Philodemus are with no case drawing us to the contrary.

It’s also inconceivable that a tree may ever have lungs or a nose, or grow hair or feathers, because never in our experience have we seen this. Cassius explains:

The entire argument of the whole book rests on “can our senses provide us an example on which to rest our argument” versus the Stoics “our argument can rest on logic through syllogism even though we have never seen or touched or sensed an example of what we are talking about”.

In my opinion, what the Epicureans were saying is that always, unless you can point to evidence of the senses as proving that each part of the syllogism is true, it is “inconceivable” that the syllogism can be true.

And so, again, we see the insistence on evidence presented before the tribunal of our natural faculties. As to the definition of syllogism, we generally find inference, conclusion, computation, calculation. Our friend Brian says that we may define it as a collected bit of observations.

One final note must be made based on commentary in page 168 of the book, which reasonably states that “the degree of certainty of an inference is often relative to the amount of variation observable”. Ergo, in addition to instances where there are no known exceptions, there may also be instances where exceptions are few or limited to very specific conditions, which limits what can be inferred based on evidence. The book cites as an example how we can establish with certainty “the actions of certain poisons, but only relative analogies can be established on the goodness or badness of foods”.

So, to conclude our discussion on syllogism:

If the premises in a syllogism can be validated by evidence / canon, then a syllogism may be valid.

If it’s not based on sense evidence or conflicts with reality, it’s not valid.


Our friend Cassius adds that probably the correct way to say it is that any syllogism, in order to be correct, must have all premises validated by evidence. It’s not complicated. In fact, many of Philodemus’ arguments against the Stoics simply refer back to the insistence on evidence. For instance:

… anyone who infers well about the unperceived objects that accompany appearances observes carefully the manifold variety of appearances in order to be sure that there is no conflicting evidence. He considers it impossible that the nature of things and their generation from each other should be inconsistent with appearances. – On Methods of Inference, XXXIII

Since we are not extremely fond of obscure language in Epicureanism, we prefer to say that we conclude or calculate this notion from that fact, rather than make use of technical terms that almost no one uses.

In fact, in many of our discussions with logicians we end up concluding that logic oftentimes makes things more difficult and complicated than they need to be, or than they are by nature.

Examples of Why The Canon Matters

Let’s have some fun and formulate some syllogisms so that these teachings can become strong in our souls, as Epicurus used to say.

Allah is the Greatest
The angels work for Allah
Therefore the angels are not the greatest beings in the cosmos

The above syllogism is perfectly valid. It has two premises and a conclusion which are internally consistent: there is one cosmic being who is the greatest, and creatures who are minions of this being, ergo these minions are not the greatest beings, they are secondary. A theologian would be happy with this perfectly logical syllogism. For us materialists, however, the lack of evidence for the existence of both Allah and Allah’s angels represents a serious problem. The premises, and therefore the conclusion, must all be false unless and until we can corroborate all three.

As we said earlier in these reasonings, Epicureans favor the method of similarity where we infer, for instance, from a finite sample to a whole population, or from the behavior of microscopic bodies to atoms or to bodies in heaven. According to James Allen, “if the opponents are right, the method will mistakenly project features belonging to the items in our experience onto items outside of our experience to which they do not belong“. He then goes on to cite that atoms would have color because things in our experience do, and that we would exclude possibilities that exist because they’re not in our experience.

However, we should answer to this by use of one final example, which is of particular interest because it confronts us with what happens when there is a unique exception: the platypus, an aquatic mammal who lays eggs and has a bird-like beak.

No mammals lay eggs
The platypus is a mammal
Therefore the platypus does not lay eggs

The above syllogism is valid, and if we did not pay importance to evidence, we would be obliged to conclude that the platypus does not ley eggs … except that it does. Evidence, not syllogisms or logical games of any kind, produces accurate conclusions about matters both rare and common. Prior to the discovery of the platypus, the syllogism would have been thought accurate. Now, not so; and it’s only because of evidence that we are forced to expand the definition of a mammal, and never from syllogisms. This is why the canon matters: it is our connection to reality.

Further Reading:

 Philodemus’ On methods of inference: a study in ancient empiricism
Inference from Signs: Ancient Debates about the Nature of Evidence, by James Allen
Against the Vulcans

Back to the Main Page

Tweet This