Thomas Huxley, Darwin's famous 'bulldog' advocate for the theory of evolution, wrote in an 1860 letter that in science one should “Sit down before fact like a little child, and be prepared to give up every preconceived notion, follow humbly wherever and to whatever abyss Nature leads or you shall learn nothing.” He wrote this in a context of a discussion (by letter) about religion and related topics with his friend and theologian Charles Kingsley.
This represents Huxley's view about theology and empiricism. Essentially, he's advising to confront Nature with an open mind, and see what lessons one can learn; don't accept dogma, be humble before the facts.
Science should be the area of human affairs that, above all, is humble before the facts. But in many ways science is all too human for that. Although very empirical, scientists tend like any other group to accept, defend, and stick to a canon of dogma which only changes under the force of substantial new evidence, or pressure from a rival camp.
So, for example, when aspects of today's genomics are questioned, the typical response is that, sure, we don't know everything, but if we can just acquire more data (usually meaning more Big Data) the uncertainties will become clear. It is a view that the fault, dear Brutus, is in our sample size and technology, not in ourselves.
We often feel that while more data, or more forms of data than we currently have, will lead to some new discoveries, and plenty of incremental new knowledge, it won't automatically yield deeper insights. Rather than being so cock-sure that we have the right theory but only insufficient measurement technologies, we tend not to think that there may be wholly unsuspected factors that could account for things not currently being explained very well.
We tend to call such possibilities "dark matter", in quotes because we're not suggesting that the dark matter of current physics is (necessarily) a missing ingredient that could turn our current theory every which way, as in other major changes in scientific view. We have no particular insight, and have no way to know whether any such transformational theory is possible. Perhaps the complexity with which we try to wrestle these days really is as complex as it seems, and no new factor will change that..
But there are, occasionally, discoveries that show us how deep convictions can be upset by new findings. An example that has actually been known for some time, but which we only learned of recently, via BBC Radio 4's 'The Life Scientific' interviewing Oxford's Professor of Circadian Neuroscience Russell Foster, is our 'third' form of vision.
Everyone knew of light-sensing and color-sensing pigments, called opsins, that are coded by genes and expressed in our retinas. Likewise, the pineal gland has other light sensing and time-cycle rhythm monitoring functions. It isn't directly exposed to light in us or other mammals (that we know of), but it does respond to neural signals of light perception sent from the retina.
Blind people were long thought to have no light perception, because they had no rod or cone cells in the retina or for other related problems these cells did not transmit reception signals to the brain (including the pineal gland). The idea that blind people could not sense light was very strong dogma and led for example, to the replacement of real eyes with glass eyes in blind people (sometimes for legitimate reasons such as infection of a blind eye), but generally because 'of course' the natural eye was of no use.
But various investigators, especially including Russell Foster, noticed various facts, such as that blind people could sense and maintain diurnal (day/night) cycles. Controlled experiments were done that showed that while not being able to 'see' as the vernacular and formal theory described that sense, blind people often could report with very high accuracy whether the lights were on or off. Such findings were received with dismissive hostility until they became clear enough that the older dogma simply had to be adapted.
Since then, a gene coding the protein melanopsin has been found, shown to be expressed in neural cells in the eye. Melanopsin acts as a cell receptor for light. It does not participate as retinal cells do in a pattern-recognizing field (so far as we happen to know, at least). But areas of the brain not involved in image recognition do perceive and 'understand' the incoming information.
This discovery was not a transformative one in terms of genetics or evolution. Once discovered, melanopsin and relevant mechanisms have been found to have a long evolutionary history and so on. It's an ordinary gene, and no new basic principles of genetic causation or mechanism are involved.
The point here is that this is a rather mundane example of a finding, even within current working ideas about genomics and evolution, that was totally unexpected and whose implications were rejected because there was too much acceptance of dogma.
We have to have accepted theory from which to work, and we have to resist random or wild-seeming ideas, which are always being proposed. Investigators design studies based on theories they have about life generally or their particular topic. Balancing the need for a framework for daily activity, and to remain humble before the facts is not easy in science. Scientists may be better at self-challenge than religious or political ideologues, but we're not as good at it as we like to think. It's a lesson we need to learn on a regular basis.