The intuitive view of the world is that there are substances and properties. Every property, e.g., redness, must be a property of a substance. The redness inheres in the substance of the red apple.
This simple observation leads us to enumerate properties and substances. Since assertions cannot be flightless, we might suppose that assertions are made of a different substance than are birds. We are also led to consider whether bare substances can exist, i.e., whether it makes sense to consider substances absent of any properties.
Needless to say, the whole subject soon gets rather complicated. There are many philosophers who doubt that bare substances make any sense at all. They ask whether an apple substance is an apple at all if it has no properties of an apple. For if it is not, there may be no substances at all. Apples may be just bundles of the properties that define them.
As a study of the way language and intuition work, this is all harmless fun. However, the ship of substance theory runs aground when it makes the leap from what we experience in our study of substances and properties to metaphysical claims about kinds of substances. For example, we know that there are properties that inhere in minds that do not inhere in rocks or in thermodynamic heat. Hence, the metaphysician is tempted to say that minds are made of metaphysically different stuff from rocks and heat.
One argument the metaphysician may employ for making this distinction is that rocks and heat are both detected with our five senses, while propositions are detected only mentally. In other words, it is suggested that we should cleave substances into different categories according to the kinds of experience of their properties.
The problem with this claim is that, a priori, "the five senses" are an arbitrary conglomeration. They are not a unified whole, but are a collection of independent senses. Why should our sense of touch be grouped with our sense of vision?
The best explanation for this grouping is history of correlation. We are accustomed to thinking that the donut we feel in the palm of our hand is not something other than the donut we see sitting in the palm of our hand. The correlation of these senses is strong, and prior to any philosophical consideration. The five senses are sewn into an integrated model by our brains when we are young, and that's why we group them so naturally. But we don't have rigorous philosophical grounds for making this grouping. In principle, the donut we feel in our hand may be something different from the donut we see with our eyes.
Of course, no one doubts that the two (or five) donuts are one and the same thing, or that a donut has properties of taste, shape, weight, smell and color. Yet, donuts are not elementary sensations. There is no donut sense independent of our five standard senses (sorry, Homer!). Donuts are correlations of all of their properties.
Hence, we are not justified in establishing metaphysical classes of substance on grounds of method of experience of their properties unless we're willing to create new classes for sight, hearing, taste, touch and the like. And since we generally regard it as sensible to think of donuts as things of a single substance, we can't consistently choose to create separate substances for each kind of property experience.
This means that mental objects such as logical contradictions, feelings, moral dilemmas and propositions should not be regarded as being made out of different stuff just because the properties of these mental objects do not directly involve the five senses. Nor, for that matter, is it reasonable to consider all mental objects to be of the same substance.
The clincher is the fact that mental sensations are strongly correlated with physical ones. This is seen in brain scans (e.g., fMRI) and in studies of patients with brain injuries. Today, almost every kind of mental sensation has been correlated with physical functions in the brain. If we are to resist reducing mind function to brain function, we should also resist reducing water to H2O.
As I said, it's not that we cannot analyze language and intuition in terms of categories, it's just that we are not justified in claiming that things cleave into metaphysical categories of being based on these intuitions.