It is helpful to distinguish between what I shall call deep common sense (DCS) and local common sense (LCS). DCS is the kind of common sense that all humans share and which has existed for hundreds and thousands of years (and may well have been present in earlier hominids). It is a human universal, independent of time, place, and cultural context. It is deep in the sense that it underlies other aspects of what we call common sense, such as local knowledge of places and things: the most fundamental stratum of human belief or commitment. LCS is what gets added to DCS in the way of local and specific beliefs, whether folk or scientific—it puts flesh on the bones. DCS is analogous to universal grammar, as Chomsky understands it, and LCS is analogous to particular spoken languages.
What does DCS contain? Answering this question fully would require a lengthy conceptual and empirical inquiry; I shall offer the outlines of an answer, focused specifically on the question of our deep beliefs regarding the external world. The most basic component of DCS is the belief (or assumption) that we enjoy a range of experiences: perceptual, emotional, imaginative, cognitive, dream, and bodily. These are experiences that a subject of consciousness possesses—and we take this subject to exist from moment to moment. We also assume that space and time form the background to experience: space and time are objectively real and we exist within space and time. We accept too that our experiences correspond in some fashion to objects within space and time—objects that exist independently of our experience and are responsible for our experience. Thus part of DCS is belief in an external reality—a world beyond the experiencing mind. But it does not follow that we believe in “material objects” or that our experience mirrors or resembles the nature of external objects. I want to suggest that DCS is neutral on such questions: it holds merely that external objects exist in space and time, but it takes no stand on what these objects are in themselves or whether we perceive them in their objective nature. That is a matter for LCS: are they collocations of material particles, ideas in the mind of God, permanent possibilities of sensation, windowless monads, centers of sentience, or what have you? Are they directly perceived or indirectly represented? Do they have all the properties they appear to have? This kind of belief can vary from group to group, unlike the universal assumption that there are external objects of some sort that stand in some relation to human experience. DCS is schematic regarding the nature of the external world: it is a placeholder for more specific beliefs, both folk and scientific. But it is committed to the notion that there is an external world. It is not similarly committed to the existence of other minds: an individual might not have anyone around her possessing a mind (she is entirely solitary); and anyway it is not mandatory to accept that other people (or animals) have minds. That again is part of LCS: it is not part of our deep commitments about reality—any more than a belief in trees and dogs. It is not part of the very structure of the human conceptual scheme. We have no deep belief that insects and reptiles have minds (even though they probably do) and human minds don’t differ as a matter of principle: this is an optional belief so far as DCS is concerned.  But we are deeply committed to the idea that we live in a spatiotemporal world containing objects external to our minds.
Anything else? I think we are committed to belief in logical and natural laws: we hold that thought is governed by rules of reasoning and that nature is subject to lawful regularities. What these laws precisely are lies outside the scope of DCS: that is a matter for research and discovery, and people will vary in their answers. What we deeply believe is just that logical and natural laws exist. Even our distant prehistoric ancestors believed in valid reasoning and predictability via law (they weren’t stupid). Likewise they believed in the moral distinction between right and wrong, and also that this distinction matters. By no means is DCS cautiously empiricist or cowed by skepticism: it has no trouble accepting entities that go beyond immediate experience—space, time, the self, external objects, logical and natural necessity, moral norms. DCS conceives of human experience as existing within an objective framework, not as constituting the whole of reality. Experience is internal to this framework, not the framework itself; it exists alongside non-experiential things. It is just that it is neutral as to the nature of the things to which it is committed, except in a schematic way. It is realist but unspecific.
What DCS does not contain is belief in such things as the sun, the earth, mountains, oceans, elephants, trees, deserts, ancient civilizations, democracy, gods, creationism, and so on. We may describe these as part of the common sense of particular groups, but they are not part of the universal structure of beliefs embedded deep in the human mind. DCS does not even contain a belief in material objects in anything like the sense this phrase has in LCS: the very concept of matter is alien to DCS, being a concept of science and its cultural context. Does a dog believe in the existence of material objects—does it bring the objects of its perception under such a concept? No, though it no doubt assumes that what it sees exists independently of its seeing it. You can believe that rocks and streams are reincarnations of the souls of the dead and still subscribe to DCS. Possibly our remote ancestors did have just such local cultural beliefs as a way of fleshing out their schematic conception of reality. The contemporary idea of material objects is heavily imbued with scientific and philosophical theory, not a primitive component of common sense at its most fundamental level. Nor does DCS assume that everything that belongs to appearance belongs to reality: it doesn’t assume, say, that external objects are objectively colored. It understands the difference between appearance and reality, thus allowing for the possibility that things may not be just as they seem. Naïve realism is not part of its portfolio: that is a matter of local theory not core commitment.
This two-tier model of common sense enables us to say something it is important to say: while local common sense has been challenged by the discoveries of science, deep common sense has never been so challenged. For example, the geocentric theory of the universe has been refuted by science, which is surely part of the LCS of many groups, but science has never challenged the belief that there are external objects, which is a core tenet of DCS. Nor has science undermined the belief in space and time, though particular folk theories of space and time have no doubt been challenged. Philosophers might question the central tenets of DCS, but scientists have not; so we cannot say that science has ever falsified DCS. Nor is it easy to see how science could falsify DCS: it is, if a theory at all, a theory at a more abstract level than the level reachable by empirical science. How could science ever prove that space and time are unreal, or that there are no objects external to the mind? How could it prove that there are no logical or natural laws? DCS is not vulnerable to scientific refutation, though LCS certainly is.
There has been a tendency to suppose that if some part of common sense has been undermined by science, then any part may be; but that is a non sequitur, since the parts vary in their content, as with DCS and LCS. And the stratum occupied by DCS is quite substantial—by no means trivial. Thus the core of common sense is invulnerable to refutation by science (though philosophy might be able to make dents in it). DCS is commendably cautious when it comes to the specific nature of the external world and our relationship to it, though LCS is susceptible to error and absurdity. It is tempting to see DCS as innate and ancient, deriving in large measure from non-human animals, while LCS reflects passing cultural influences, fads, and fashions. It is as if DCS says: “We really don’t know much about the nature of the world beyond the mind, so let’s not commit ourselves on the subject; suffice it to note that there is an external world of some description”. And it is true that nothing in immediate experience offers a clear message about the objective nature of the external world, since it is compatible with many possibilities—experience might be massively misleading and it doesn’t answer a host of questions about the nature of reality. Things may not be what they seem and they may have a being that transcends their seeming—DCS accepts those truisms. It takes LCS to venture beyond this kind of minimalism, accepting things like naïve realism or the atomic theory. Thus what is universal to human belief survives challenge by science, while what is local can readily be so challenged. In this sense, common sense has not, and probably cannot, be undermined by science. What is called common sense is often proto-science, and hence can be refuted by superior science, but not all of common sense counts as proto-science—specifically, the bare belief in an external world. 
 Someone who resolutely denies that that other people have minds, based on a theory of his own invention, is rightly regarded as eccentric, but he does not reject a universal tenet of common sense—while someone who thinks that he is the only existent thing does violate basic common sense.
 It is as if deep common sense has been designed to fit a variety of possible environments or worlds: it studiously avoids too much specificity. It is then filled out by particular belief systems that may or may not survive scrutiny. Compare: all human languages contain verb phrases and noun phrases, but they vary in the specific phrases they contain.