It could be said that if a foundationalist believes that “basic beliefs are about observable physical objects” (e.g., “there’s a dog”), then he's not in fact a genuine foundationalist at all. If he believes that a dog is in fact a dog, then he has already brought beliefs to bear on the dog-experience to start with.
How does he know that it is in fact a dog? Because of his prior beliefs about dogs. Perhaps he tacitly uses the conditional: If an object or animal barks and has four legs and a tail, then that object must be a dog.
“There’s a dog” can't be a foundational belief in any sense of that term. The only thing he could say is that it is a belief caused by actual interactions with dogs, whereas some of his other dog-thoughts aren't caused by causal interaction with concrete dogs. In that case, perhaps we could say that the causal interaction with a dog is foundational; though not the belief itself. In that case, as Susan Haak says, it is
“only propositions, not events [or objects], that can stand in logical relations to other propositions [or beliefs]”. (227)
The causal interaction itself is neither a belief nor a proposition. Therefore it can't “stand in logical relations to other propositions” or beliefs. The causal interaction with the dog might have been (as it were) a first cause of the belief or beliefs about the dog; though in and of itself, it is neither evidence for such beliefs nor a justification for further beliefs.
In any case, the same causal context - taken only in itself - can cause different beliefs in different people and possibly different beliefs in the same person at different times. The interpretations of our causal contacts depend on our prior beliefs and the prior concepts which we apply to our causal interactions or contacts. And even if a particular causal contact brings about the formulation of new beliefs or new concepts, these will still be dependent upon or related to prior beliefs and prior concepts. Such things do not spring up ex nihilo and neither are they “first beginnings” or “first causes” (as it were).
Even the adverbial versions of foundationalism smuggle in prior concepts and beliefs.
For example, “I am appeared to brownly” still uses the concept [brown]. And even the concept [appears] implies that the speaker is making a distinction between appearance and reality. In addition, even if that statement isn't object-involving, it's still property-involving. And where there are properties, there are also objects to which the properties can be predicated. Unless the subject thought that brown itself, or brownness itself, qua trope, appeared to him “brownly”. That isn't likely.
If the speaker can identify the property- concept [brown], then why can’t he also identify an object-concept, say, [dog]? He might not have done so that this particular time, of course; though there was nothing to stop him doing so. And if something “looks like a dog”, it still looks like a dog even if it isn't an actual dog. The prior concept [dog] is still (surreptitiously or tacitly) used by the adverbial foundationalist.