A brief epistemic history of Western thought (part 3): Science as a method

Science gives us methods to trust, rather than intellectual systems. Does this provide epistemic certainty?

Today, science and scientific practices so thoroughly dominate the way we think that it is hard to imagine them as anything other than natural and inevitable. However, what we now call science is the result of a particular set of philosophical principles refined over centuries of practice. As noted in the first ‘brief epistemic history’, it is one intellectual tradition that emerged in the 1500s and 1600s in reaction to the collapse of the Medieval Synthesis and the intellectual certainty that had provided.

The history of this intellectual tradition is harder to summarise than the explicitly philosophical tradition focused on knowledge and reality we looked at in our second epistemic history. It evolved more by practice than through theoretical considerations and has, over time, split into a range of distinct practices that we now consider separate scientific disciplines. For the purposes of a brief epistemic history, we will just focus on a few thinkers about scientific methods and epistemology - a history of the philosophy of science rather than science itself.

In the two previous histories, we encountered a common intellectual pattern: thinkers would attempt to establish epistemic certainty through a new system or way of reasoning, possibly within some boundaries, but this certainty would inevitably collapse - often into skepticism. In response, others would try to re-establish certainty and the cycle would repeat.

Scientific history is somewhat different as it begins with, and is founded on, an acknowledgement of epistemic humility: knowledge is hard and humans often get it wrong - so we need to be careful and rigorously test our findings. However, this has not meant dreams of certainty were abandoned. The focus shifted from trying to develop conclusions or systems that provide us with certainty, to methods that can provide us - over time - with secure knowledge. However, in common with the other fields, while science has clearly been successful in delivering knowledge, there are ongoing debates about the certainty that scientific methods can deliver.

Beginnings in natural philosophy

Francis Bacon is generally considered to be the Father of Modern Science, with his key ideas set out in the Novum Organum. He begins this by explicitly rejecting a key assumption of most Greek and Medieval Philosophy: that human reason is a robust and reliable instrument for generating knowledge. Bacon argued that, instead of human reason being sufficient to establish reliable knowledge, it is instead dominated by four ‘Idols of the Mind' that cause it to go astray. These Idols, of the Tribe, the Cave, the Marketplace and the Theatre, are descriptions of what we would now call cognitive biases. As such they provide a clear description of why our instinctive, rational views of the world are often mistaken. Hence human reason itself is not trustworthy.

Instead of relying on reason alone, Bacon argued for a method of careful and meticulous observation of the world; one “which by slow and faithful toil gathers information from things and brings it into understanding". He advocated for an inductive approach, where we carefully establish patterns in observations and then make appropriate generalisations. In his thinking, this was more rigorous and would generate more secure knowledge than the previously dominant Aristotelian approach. This, although also grounded on observation, typically jumped to a generalisation once an underlying explanation could be found, rather than waiting to observe consistent patterns. For Bacon, we couldn't immediately trust human reason and so needed to be more careful and rigorous.

Around the same time as Francis Bacon, Galileo Galilei produced some pioneering works of natural philosophy that set directions scientific progress would follow. Like Bacon, Galileo placed a very strong emphasis on the importance of experiments and testing ideas carefully through observations. He therefore also rejected the sufficiency of human reason for establishing knowledge.

Galileo also rejected the previous consensus in three other directions. For one, he emphasised mechanistic descriptions of physical events - in contrast with both Aristotelian teleological explanations and religious or spiritual explanations. Secondly, he explicitly tried to capture broad scientific theories in a small number of universal, abstract principles that applied to all matter. And thirdly, he relied on mathematics for his mechanistic explanations.

It is worth noting how shocking some of these methods would have been to Ancient thinkers, especially the reliance on mathematics. In the Ancient Greek, and Medieval, worldview, mathematics was an exploration of perfect logical concepts. For them, the heavens were uncorrupted and therefore mathematics could be used to describe heavenly things like the orbit of planets. However, our physical world was deeply corrupted and therefore mathematical explanations could never be relied on here. The idea that the physical world was intelligible and adhered to predictable laws was new and, for these thinkers, largely grounded in their religious convictions.

In this context, it should be noted that Galileo's famous argument with the Catholic Church of the day was not simply about whether the Earth revolved around the Sun, or vice versa. That issue was the flash point for a deeper clash of epistemology - whether the unified philosophico-religious framework of the previous millenium, with its reliance on the sufficiency of reason and particular revealed authorities, was genuinely open to challenge. Galileo’s science challenged an entire established worldview that adherents thought was epistemically certain and unchallengeable. In doing so, he was a threat to the intellectual foundations of a whole social, religious and political order.

Induction or Hypotheses

While their emphases were similar, there were some key differences between the methods of Bacon and Galileo, which have echoed down the history of science. Bacon emphasised a careful inductive approach where conclusions could only be drawn from, and then tested by, careful observations. This strong emphasis on induction was followed by, among others, Isaac Newton and JS Mill. Galileo, by contrast, was more interested in developing general principles and theories which needed to be consistent with, and tested against, observations. This approached, followed by thinkers such as Gottfried Leibniz and William Whewell, reflects an approach now broadly labelled the hypothetico-deductive model.

To understand the differences, it is helpful to examine Newton’s attitudes to his scientific practice, especially as he strongly asserted that 'I don't make hypotheses'. Newton broadly followed Bacon in the view that close observation and careful induction were necessary to ensure that natural philosophy provided true or trustworthy conclusions. Extending conclusions beyond the evidence, as he thought the act of formulating hypotheses did, was skating on thin ice. For Newton, this provided effective epistemic certainty:

In experimental philosophy, propositions gathered from phenomena by induction should be taken to be either exactly or very nearly true notwithstanding any contrary hypotheses, until yet other phenomena make such propositions either more exact or liable to exceptions. This rule should be followed so that arguments based on induction may not be nullified by hypotheses.

Notably, Newton took this approach seriously enough that he refused to come to conclusions on issues, such as whether light was a particle or a wave, where the observations didn’t justify a conclusion.

However, to compress a long story, two significant problems arose with careful inductive approaches similar to those advocated by Bacon and Newton. As a matter of history, what we now call Newtonian mechanics and Newton's Laws of Motion were not broadly accepted as true simply from Newton's careful induction. Instead, they were treated as hypotheses that weren’t fully accepted until there was experimental verification for a range of consequences of his conclusions over the decades after his death.

The second problem were doubts about the logical rigour behind an inductive approach. This was most forcefully put by David Hume during the 1740s. Hume argued that there is no inherent logical reason to expect that a pattern we have seen in the past will continue: why does something that has happened in the past have to happen in the future? Hume argued that we assume inductive patterns will continue simply via custom or habit - hardly a rigorous basis for epistemic certainty.

Variations on this critique have continued through to today, and pose an ongoing challenge to the epistemic certainty scientific methods provide. While some, like JS Mill, tried to modify pure inductive approaches, following Hume it slowly became clear that a simple inductive argument could not establish scientific knowledge with any logical certainty. This was extended by various arguments in the twentieth century that there are, in fact, many different generalisations that are equally consistent with a given set of observations - so induction can’t establish any logically rigorous conclusions. We generally select one for simplicity or explanatory power, but these are considerations outside of pure induction.

Science as theory

By the end of the nineteenth century, it was increasingly clear that scientific progress was being made by developing broad scale hypotheses, or more commonly theories, that had explanatory power, and then testing these against reality via experiments. Good evidence of that is in the way that much established nineteenth century science is still described in these terms - germ theory in medicine, the theory of evolution in biology or atomic theory in chemistry. Moreover, following Galileo’s lead centuries before, science increasingly progressed by discovering broadly applicable mathematical equations that were presumed to hold generally, such as Maxwell’s Equations of Electro-magnetism or the Laws of Thermodynamics.

The practical successes of this approach raised questions of epistemic warrant: to what extent could we be certain that these theories were true as descriptions of the real world? We couldn't really test them directly (e.g. no-one could easily see an atom or a germ) but only that the consequences of them held up. Philosophy and theories of science have been grappling with this ever since.¹ Broadly speaking, the questions have only got more difficult.

For example, Pierre Duhem added an extra wrinkle to these concerns with some observations in the early 1900s. To pick an example used by Duhem, if we are conducting an experiment that uses a thermometer, we need to be aware that the measurement tool itself embodies a theoretical understanding of the world and its reliability depends on that underlying theory. A thermometer doesn't tell you the temperature directly but (in those days) relied on the thermal properties of mercury expanding in a regular way with temperature so lines on a board corresponded with genuine temperatures. If there was some problem with the science governing the expansion of mercury, then the measurements from the thermometer would be (sometimes) incorrect. This point holds for almost any non-trivial piece of equipment used in a scientific experiment.

As Duhem pointed out, if an experiment failed, how could we actually know that the failure was in the theory we were trying to test or in one of the various other theories necessary for the equipment we were using? In other words, how can we be sure what theories our experiments are testing and, therefore, what knowledge they give us?

In a different direction, Karl Popper drew on a worry implicit in inductive approaches to science to argue that science gave us little epistemic confidence in the truth of our theories. Newton (above) had noted that scientific conclusions are only true 'until yet another phenomena' changed our conclusions. Popper extended this idea to argue that it is always possible that a current scientific theory will be disproven by future experiments. This means they can never be definitely shown to be true (or confirmed). For Popper, scientific theories can only be falsified - shown to be false - and our best current science are those surviving theories that have not been falsified. The best we can get is not false rather than true.

While this strikes many as overly pessimistic - surely we can confirm at least some theories - it also lead to an interesting corollary. Popper also argued that only theories that could (at least in theory) be falsified can qualify as scientific. If there is a theory that can fit all possible evidence, is it really a scientific theory worth pursuing? For Popper, the only scientific theories are those that say something meaningful about the world and therefore could be proven wrong.

Science in practice

From the mid-twentieth century onwards, scholarship on the nature of science has diversified but generally focused much more closely on what scientists do in practice. In various ways, this has continued to undermine any universal scientific claims to epistemic confidence or certainty.

For example, Thomas Kuhn argued from historical examples that scientists tend to work within strong paradigms, or universalising theories, and the majority seek only to extend or explore a particular paradigm. They do not consider alternatives and only give one up in favour of another when the first collapses as a viable explanation. On his account, a scientific theory being considered true really just means that a majority of (powerful) scientists in the relevant community believe it. Should that give us epistemic confidence in science or not?

Bruno Latour, in a different direction, pioneered the anthrophological study of scientists and found there are a wide range of learned skills, instinctive judgement, social behaviours and cultural idiosyncracies. Many of these were surprising to the scientists themselves and difficult to justify. Scientists weren’t the dispassionate and purely rational pursuers of truth that many thought they were. This does not mean their conclusions are wrong, but opens up questions about what our epistemic confidence in science is grounded in.

More recently, as researchers have explored the actual practices of scientists, it has become apparent that there is less of a universal scientific method and more a collection of practices that have been grouped together as scientific. These differences between methods only deepen the questions about why science provides us with knowledge and whether there is a scientific method that deserves our epistemic confidence.

Is there any scientific certainty?

Once again, an important Western intellectual tradition that seeks, or claims, epistemic certainty yet the various attempts to justify this certainty come apart when we look closer. The scientific tradition is notable, as unlike others we have looked at, it seeks to build epistemic confidence through a method that assumes fallibility in human rationality. However, the exact nature of the scientific method is slippery to pin down and open to doubts.

This leaves us with something of a paradox. Science as a method, or a set of practices, has clearly been highly successful at generating knowledge of the world and has improved our lives in many different ways. However, we can’t convincingly explain why it has been successful and don’t have clear rationale for trusting it gives us reliable knowledge - beyond the fact that it has worked. We are confident science gives us knowledge but we can’t really say why.

Perhaps, however, we have been looking at it the wrong way round. Science, as a tradition, was founded with a strong epistemic humility: human reasoning was liable to error and therefore we needed to test thinking rigorously against physical evidence. This humility demands that our focus is on the world around us and how it works, not on our logic, intuitive methods, thought patterns or expectations. This has meant that scientific methods and theories have, slowly over time, come to better reflect the nature of the world we live in. Methods for different parts of science have diverged as the focus of disciplines spread and scientists learnt how to make sense of different aspects of the world.

Assuming there is a single universal scientific method that was guaranteed to produce knowledge presumes there is some aspect of human rationality that is always epistemicly reliable. But this is at odds with science’s founding principle: epistemic humility. Perhaps the success of science is less a result of any method and more due to this fundamental attitude. The ongoing focus on submitting human ideas to the discipline of physical testing, and rejecting great ideas that don’t work, has built a reliable body of knowledge because we trusted human rationality less.

1. For the sake of simplicity, this article is going to leave a large amount of more technical, logic focused discussions of the nature of scientific knowledge. These mostly cover the same themes, just more precisely.