As a graduate student, I was fortunate enough to be primarily trained by one of the outstanding archaeologists of the previous generation, Dr. Michael Glassow*. Dr. Glassow had first risen to prominence in the 1970s, when a school of archaeology known as "processualism" (because it was focused on the "processes" of culture formation, the activities that form a supposed "universal background language" of how cultures change and function, with specific cultures being looked at for both data and as case studies).
One important aspect of Processualist archaeology, and one that has been criticized by the various schools of "post-processualist" archaeology, is a tendency to develop models of how people "should" behave in order to both create predictive models of culture, and also to provide a baseline against which variations could be examined - basically, if someone is not doing what is optimal, then we can look to see what causes this anomaly.
A typical model might calculate the nutritious value of a food and also the difficulty in obtaining said food, and then work a hierarchy of preferred foods for a given group of people within a given area, with the highest nutrition-to-work ratio being labelled the "favored" or "top ranked" food. Often, this works. Archaeologically, the "top ranked" food is the one with the most abundant archaeological signature (either by way of faunal remains, or evidence of the food being important to tool-making or the settlement pattern, etc.). These models provide a good framework for archaeological studies. They work pretty well most of the time, but, often enough to be worrying, these models provide approximations of the real world, but fail to exactly describe real-world situations - rather than being anomalies, people not behaving in an optimal manner is a rather mundane thing.
Case in point: historically, there have been trout runs in the Santa Ynez Valley in southern California**. Given the high nutrient value of fish meat, these should have been eaten in large numbers by the prehistoric peoples of the Santa Ynez Valley. However, when you actually look at the archaeological sites, the bones of these fish make up less than 1% of the identified fish bone in Santa Ynez Valley sites, and less than 1/10 of a percent of the total meat weight consumed at these sites. However, marine fish - fish imported from the coast and carried over a mountain range into the valley - make up more than 99% of the fish bone identified at these Santa Ynez Valley sites.
What the Hell? People weren't willing to eat the fish available locally, but would travel considerable distance, or trade over considerable distance, to get fish from the coast?
If you point this out to strict adherents to the nutrient/work model, they will provide a wide range of explanations which range from the unlikely (the fish must not have been available prehistorically - despite the fact that there is no known ecological reason to conclude this) to the just plain weird (dogs would eat the bones of the trout, but no other fish in the sites!). None seem to want to accept that the evidence indicates that people were simply not using a high-ranked resource. Strange, but true. Less strict adherents, those like myself who think that the models are valuable but not the be-all and end-all of cultural explanation, will accept what the evidence indicates, and begin looking for possible explanations to tell us why the archaeological record doesn't match the model.
There are many cultural mechanisms that might explain why these fish didn't make it into the sites - religious taboos against eating the fish, agreement between valley peoples to leave this resource alone except in dire situations, etc. - all of which seem unlikely, to be certain, but all of which can account for the archaeological evidence in a way that simple models of nutrition and labor can not.
There is an even simpler explanation, one that most archaeologists actually tend towards, but one which nonetheless remains controversial anytime that it is explicitly brought up in publication or discussion (although it is usually uncontroversial when it is merely implied rather than being outright stated). That explanation is simply that nobody is omniscient regarding their environment. People lack important information, often have misinformation, might not be inclined to gain information, and may have cognitive or cultural biases preventing them from gathering accurate information.
This isn't such a strange idea. Look at people in the modern world - there are starving people who will not eat perfectly good foods that their religion holds are unclean, or with which they are not familiar (and therefore unaware of the benefits of or else consider "gross"). Likewise, ethnogrpahers have found that while modern hunter gatherers have an amazingly complex understanding of their environments, they don't know everything about it (any more than a modern urban person knows everything about their neighborhood - though hunter-gatherers do tend to know more about their surroundings than us urban folks do, it is often a matter of survival after all), and sometimes a lack of information that seems on the surface inconsequential can seriously impact the resources utilized.
Basically, the processual models require that the people generally be purely rational (not allowing superstitions to alter their actions - which is unrealistic for any group of people at any point in human history), be motivated purely (or at least principally) by the desire to gather resources (usually food, though other resources may) while reducing expended effort, and to have had not just a thorough knowledge of their surroundings but a complete knowledge of their surroundings. These models also require that the model-maker take into account all information about the physical environment and physical needs of the people being studied. These conditions, of course, are not realistic.
Which is not to say that these models are not useful. They give us an idea of what people should do under "ideal" circumstances, and by looking at how any given culture varied away from the model, we can get a sense of where to look for the ways in which circumstances will invariably not have been ideal, and from there start putting some flesh on the cultural model that has been developed.
*No disrespect should be construed towards the other people who trained me, foremost my MA committee: John Johnson, Michael Jochim, and Phil Walker, all amazing anthropologists as well, and also the general archaeology faculty: Kathy Schreiber, Brian Fagan, and Stuart Smith. All of these people were well-respected, and with good reason, but Professor Glassow is the one who I worked with the most.
**Yeah, yeah, the locals claim that it's the "central coast", but this is simply because they've never bothered to actually look at a map.