Mental content is what thoughts are about. This seems like a simple enough idea, but if it really was as simple as it seemed I would have nothing to write about tonight, and there would be a lot of blank pages in the philosophy journals. One problem with mental content is whether that content is can be defined (pinned down) in terms of internal factors, or whether we must make some reference to the external world when describing mental content. Another problem is whether all mental content can be formalized with propositions or whether there are some thoughts, specifically sensations, that are non-conceptual. I am not going to address the second problem here. Instead I will assume that all mental content can be reduced to some set of propositions. If it turns out that not all mental content can be formalized, which seems likely, the account given here might be extended to cover the other cases, or such an may not be necessary (possibly the internalism / externalism problem does not arise for these kinds of mental content). But I am not concerned with that detail at the moment simply because the internalism/externalism question revolves primarily around the cases that can be formalized.
I think everyone’s first intuition is that mental content must be definable in terms of internal factors. After all we certainly know what our mental content is, and all we have access to are the internal factors. But then we consider cases such as virtual worlds. In a virtual world an inhabitant may think that the object in front of them is a dog, which they would formalize with the proposition D(a). But we know that they are really in a virtual world, so the object in front of them is not really a dog, but a virtual dog, and so their mental content is really best formalized as VD(a), since given the world they live in when they think about dogs they are really thinking about virtual dogs, even though they don’t think of them as virtual dogs. And some people are won over to externalism by arguments such as these. But then we consider that a person could, in their sleep, be transported to an identical virtual world and that they might never know it. In this case their mental content would have changed without their knowledge, as now when they see a dog, and think that it is a dog, we can best formalize their thought as about a virtual dog. And this implies that mental content can change without the knowledge of the thinker, and thus that the mental content is largely independent of the thinker, which seems preposterous. And so some people are won over back to internalism.
It is not the case that one side of the argument, as I have presented them above, is wrong. What is really happening is that they are discussing two different things, two different and systematic ways of formalizing mental content. I think that the internalists’ formalization best characterizes what we think about, in most cases, as mental content, but that is neither here nor there. The differences stem primarily from looking mental content from two different points of view. The internalist considers mental content as found in the mind of the thinker while the externalist considers mental content in terms of how the world really is.
The externalist usually considers formalizing mental content with some kind of causal picture in mind. To the externalist the mental content of a thought is defined, in some way, by the objects that are causally related in the right way to that thought (well, most externalists). So if our thought that something is a dog is usually caused by observing what turns out to be a real dog then it is best formalized as D(a), but if we live in a virtual world then our thoughts are normally caused by what is really a virtual dog, and hence are best formalized as VD(a). In essence the externalist formalization is assuming that since thoughts are about some objects external to us mental content is best formalized by finding the states of affairs in the world that most closely correspond to our thoughts.
In contrast, the internalist position attempts to formalize thoughts as they are found in the mind. If we think of dogs as real, independently existing creatures, that are made out of matter then the internalist says that the formalization that best characterizes this thought is D(a), no matter what is actually out there. Thus the internalists formalization admits the possibility that what we think may be radically in error, that is if we really are living in a virtual world. Now this is not to say that the internalists’ formalization isn’t really about the world. The formalization is meant to capture the worldly property of being a dog, it is just that the internalists allow that we can think of an object with certain properties even if no such object is there. For example, if we thought of heat as some kind of substance that was found in objects, then when we thought that there was heat around us we would be thinking that there was this heat-fluid around us. The externalist would say that what the thought is really about is the molecular motion that really causes heat; that it is really about the heat that we feel. And the internalist would say that is about that non-existent heat fluid, and not about the molecular motion that we are really feeling. Both are stating the mental content in terms of worldly objects, it is just that one admits the possibility that the objects in question don’t really exist.
Both formalizations have their place. However, when considering what best describes what is happening in the mind then I am inclined to say that the internalist formalization is the better choice. The internalist formalization captures the factors that affect behavior, both linguistic and non-linguistic, while the externalist formalization does not (as shown by the fact that two people with the same internal states, who behave in the same way, may have different mental contents under the externalist formalization). On the other hand some discussions of language, specifically concerning what (in the world) is meant by a sentence may require the externalist formalization, since under some theories regarding language (specifically Donnellan’s) we can refer to an object even if that object doesn’t possess the properties we think of it as having, and thus the externalist formalization best captures what we are referring to.