Philosophy of Science: What is Science and How Does It Work?

Author: Tiina Carita Rosenqvist
Category: Philosophy of Science, Epistemology, Metaphysics
Word Count: 997

Science has given us life-saving medicines, sophisticated technologies, and a much better ability to understand the world and predict the future. Because of science we now think that the Earth orbits the Sun, that water consists of molecules that are made up of two hydrogen atoms and one oxygen atom, that common colds are caused by viruses, and that our behavior is shaped by social factors.

But what is science, and how does it work? Is it unique in its approach to making sense of the world? Does it progress in a straightforward way? Does it aim to give us a truthful description of reality in its theories?^[1]

These are some core questions in the philosophy of science, which this essay introduces.^[2]

1. Science vs. Non-science

Many people think that science is our best bet at understanding the natural world. But is there something that makes science unique and distinguishes it from other attempts to understand the world, such as religion, astrology, or philosophy?

One famous candidate is falsifiability, the ability of a theory to make specific and “risky” predictions that can be tested empirically and found to be false. For example, Newton’s gravitational theory made predictions about where planets would be at specific times. These predictions were risky because if the planets weren’t there, that would have given scientists reason to think that the theory was wrong.^[3]

Scientific hypotheses are generally falsifiable in a way that the claims of, say, astrology are not. Horoscopes often make predictions so vague that they can cover almost any event (what does it mean that your love life “will take on a more complicated flavor”?).^[4]

But there are reasons to think that falsifiability doesn’t reliably distinguish science from non-science.

First, there might be scientific theories (say, in fundamental physics) that are not falsifiable in a straightforward sense.^[5] Second, non-scientific belief systems also yield falsifiable predictions. A palm reader might predict that you will have two children by the age of 30, and this prediction is easily falsified if on your 30th birthday you are childless. And while non-scientific theories might not be abandoned after such failed predictions, scientific theories aren’t always either. Newton’s theory made false predictions about the movement of Uranus. Instead of abandoning the theory, astronomers theorized that there is an unknown planet affecting its orbit, and this led to the discovery of Neptune.^[6]

Other candidates for distinguishing features also face exceptions, which suggests that there might not be a feature that all scientific theories, and only scientific theories, share.^[7]

2. Scientific Practice and Progress

People often think that science progresses steadily and linearly toward a better understanding of the world through rigorous testing of hypotheses (tentative explanatory claims) against empirical evidence obtained through observation and experiment.

But many philosophers have argued that this view is oversimplified. Some observe that individual hypotheses cannot be tested in isolation because scientists always rely on background assumptions.^[8] A failed prediction does not tell us where the error lies—in the hypothesis being tested or in the background assumptions. This idea is called holistic underdetermination.^[9]

Scientists also evaluate theories (broader explanatory frameworks) based on what they can explain. A challenge is that for any set of evidence there might be many explanations. If your evidence is wet grass, possible explanations include rain, sprinklers, and an exploded fire hydrant. The evidence alone doesn’t tell us which of these explanations is correct, or even if any of them is—an idea known as contrastive underdetermination. If the same holds in science, then a theory’s ability to explain some set of evidence might not be a good enough reason to accept it.^[10]

A response to this concern is that we often have good reasons to prefer one theory over others. Our preferred theory might be simpler, make more successful predictions, explain more observations or explain them at a deeper level, or align better with our accepted theories. These are explanatory virtues, and by comparing such virtues we might be able to rationally choose between competing theories.^[11]

That said, some have questioned whether there are genuinely objective standards that would allow us to decide between theories. Scientists operate within particular frameworks or scientific paradigms. Theories are part of these paradigms. If different paradigms have different ideas about what sorts of questions science should ask and what good answers to those questions might look like, their ideas about what counts as a “simpler” or “deeper” explanation might also vary.^[12] Still, some standards, such as predictive accuracy, could be sufficiently objective.

3. Do Scientific Theories Aim at Truth?

Many philosophers argue that scientific theories aim at truth and seek to provide us with an accurate description of reality, even when they talk about things that cannot be directly observed, such as quarks and electrons. They also think that the theories often succeed in this aim. This is known as scientific realism.^[13]

Others think that the aim is more modest: rather than aim at truth about the unobservable part of reality, scientific theories just seek to account for the phenomena we can observe. Good theories are “empirically adequate” in this way, but empirical adequacy is not a good reason to think that the theories are true. This is scientific anti-realism.^[14]

Realists think they can give a better explanation for the success of science; if scientific theories weren’t approximately true, the success of science would seem like a miracle.^[15] Anti-realists respond that the history of science shows that empirically successful theories often turn out to be false, and this gives us good reason to withhold judgement on whether our current theories are true.^[16]

4. Conclusion

Asking questions such as these is to engage philosophically with science, to try to understand it better. This is different from doing science, but philosophy and science have historically been deeply intertwined, and many philosophers of science have been scientists themselves. This is no accident. If science is to do its job well, it needs critical investigation of its starting points, theories, and methods—in other words, philosophy.^[17]

Notes

^[1] There are different kinds of sciences, including natural sciences (such as physics), social sciences (such as sociology), health sciences (such as epidemiology), and applied sciences (such as environmental science). Traditionally, philosophers of science have focused heavily on the natural sciences, and a lot of the ideas covered in this essay were initially introduced with the natural sciences in mind. That said, contemporary philosophy of science increasingly engages with disciplines such as psychology, medicine, and economics.

^[2] Philosophers of science ask many other questions as well, including questions about scientific explanation, scientific reasoning, and the role of values in science. Some of these questions arise in relation to specific sciences, such as biology, physics, or psychology. For examples, see Teleological Explanations: Purposes, Functions, and Goals in Biology by Michael Zerella, Philosophy of Space and Time: Are the Past and Future Real? by Dan Peterson, and Mindreading: Understanding Others’ Thoughts and Feelings by Emma Otterski.

^[3] The idea that falsifiability distinguishes scientific theories from non-scientific ones was introduced by Karl Popper. Popper (1962) was interested in this question because he suspected that certain theories, such as Freudian psychoanalysis, which claimed to be scientific, were better described as “pseudo-scientific.” For additional discussion, see Karl Popper and Falsificationism by Michael Zerella.

^[4] Vague claims are non-risky: they are consistent with almost any event and therefore very difficult to falsify. But vague claims can still be meaningful or valuable in some sense. For example, horoscopes could be used as a tool for self-discovery and reflection.

Falsifiability is also meant to distinguish science from disciplines that are clearly valuable but do not generally make claims that can be empirically tested. For example, mathematical claims are specific but evaluated through proofs rather than observation or experimentation.

^[5] Sean Carroll (2014) suggests that using falsifiability as a demarcation criterion for science-–a criterion that separates science from non-science—risks excluding string theory and multiverse theories in physics. He advocates for a more subtle approach and writes that “Science is (…) about explaining the world we see, developing models that fit the data. But fitting models to data is a complex and multifaceted process, involving a give-and-take between theory and experiment, as well as the gradual development of theoretical understanding in its own right.”

^[6] For additional discussion, see, e.g., Okasha 2016, 11-14.

^[7] See Hansson (2025) for some of the other candidates. Hansson points out that there is a striking degree of agreement among philosophers of science about which practices are not scientific, even if there is “almost complete disagreement on the general criteria that these judgments should presumably be based upon.”

^[8] Pierre Duhem famously wrote that “a physicist can never subject an isolated hypothesis to an experimental test but only a whole group of hypotheses” (1906). W.V.O Quine later generalized this point and argued that our beliefs form an interconnected web and that “statements about the external world face the tribunal of sense experience not individually but only as a corporate body” (1951, 38). The idea that an observation cannot show the error in a single hypothesis is also known as the “Duhem-Quine underdetermination thesis.”

^[9] This also explains why scientists often don’t just abandon their theories when predictions fail, but look for errors elsewhere in their system of beliefs and assumptions, as in the Uranus/Neptune case described in the previous section. In fact, the Uranus/Neptune example nicely illustrates holistic underdetermination: a failed prediction derived from Newton’s theory did not show that the theory itself was wrong. This gives us additional reason to be suspicious of falsifiability as the distinguishing feature of science.

^[10] For additional discussion of different forms of underdetermination, see Stanford (2023).

^[11] This form of reasoning is known as Inference to the Best Explanation or “IBE” (for discussion, see, e.g., Lipton 2017). Not everybody accepts IBE as a justified form of inference; critics deny that better explanations are more likely to be true. Philosophers have also questioned whether some purported explanatory virtues should be considered as such. For example, if the world is complex, then there may be no good reason to expect that simpler theories are more likely to be true (see Okasha 2016, Ch. 1). Others have pointed out that explanatory virtues might conflict (one theory might explain more phenomena and another might explain some phenomena at a deeper level), and there might not be a principled way to deal with such conflicts.

^[12] Thomas Kuhn (1962), a historian and philosopher of science who wrote about scientific revolutions and popularized the term “paradigm shift,” has suggested that there are no neutral standards that could be used to decisively choose between paradigms. (For additional discussion of Kuhn’s ideas, see Thomas Kuhn, Paradigm Shifts, and Academic Rifts by Michael Zerella.) Donna Haraway (1988) has drawn attention to how the social and embodied positions of scientists (their gender and other aspects of their social position) might influence what kinds of questions they ask and how they understand the world. Helen Longino (1990) has added to this by suggesting that the background assumptions that affect scientific inquiry are influenced by the larger societal context in which science operates. Haraway and Longino are both feminist philosophers of science who call for greater attention to how values, interests, and perspectives interact with scientific practice. Both suggest that genuine objectivity can be found in engaging with multiple perspectives rather than trying to eliminate perspective altogether.

^[13] There are different versions of scientific realism, but most realists are committed to the mind-independent existence of the “essential unobservables of well-established current scientific theories” and to scientific theories “mostly being right about the properties of those entities” (Dewitt 2013).

^[14] There are different versions of scientific anti-realism as well. One version states that theoretical statements that appeal to unobservables do aim at truth but should not be taken literally: talk of electrons and quarks is just shorthand for some statements about how observable phenomena behave. Another version of anti-realism states that theories like these should be taken literally but that we should suspend judgment on whether they are true. (See van Fraassen 1980.) Some philosophers have combined aspects of scientific realism with aspects of scientific anti-realism. For example, Nancy Cartwright (1983) advocates for realism about theoretical entities coupled with antirealism about theoretical laws. For additional discussion of theoretical laws in scientific knowledge, see Laws of Nature by Michael Zerella.

^[15] This is known as the “No-Miracles Argument,” often attributed to Hilary Putnam (1975). Bas van Fraassen has proposed an alternative, selectionist explanation: empirically inadequate theories are weeded out and only empirically adequate theories survive, which explains the overall success of science (1980, 40). Realists have responded that the approximate truth of the successful theories explains why they survive (see Psillos 1999, 94).

Philosophers have also appealed to underdetermination in defending antirealism. Consider the theory that there are unobservable elementary particles called quarks that are fundamental building blocks of matter. The quark theory can account for important observational data, such as patterns in particle collisions. A realist might think that this provides good reason to believe that the quark theory accurately describes reality and that quarks really exist. An antirealist can counter that, in principle, the patterns in particle collisions can be explained by alternative theories that do not posit quarks, and that we should therefore be agnostic about whether quarks really exist. A realist comeback might then involve appealing to the explanatory virtues discussed in the second section of this essay. (For additional discussion of the argument from underdetermination, see Okasha 2016, 66-70.)

^[16] For a detailed argument, see Laudan 1980. You might worry that if antirealists are right about science, then we should trust science less. Katie Morrow (2023) argues that this isn’t so. She writes that scientific practice doesn’t depend on whether realism is correct and, because of this, our trust in science shouldn’t depend on whether realism is correct either.

^[17] Philosophers’ contributions to science aren’t limited to investigations of problems in philosophy of science; philosophers have also played crucial roles in the development of scientific theories and methods of investigation (see Smith 2017).

References

Carroll, Sean (2014). “Falsifiability.” Edge.org.

Cartwright, Nancy (1983). How the Laws of Physics Lie. Oxford University Press.

Dewitt, Michael (2013). “Realism/Anti-Realism.” In Martin Curd & Stathis Psillos (eds), The Routledge Companion to Philosophy of Science. Routledge.

Duhem, Pierre (1906/1954). The Aim and Structure of Physical Theory (transl. by Philip Wiener.) Princeton University Press.

Hansson, Sven Ove (2025). “Science and Pseudo-Science.” In The Stanford Encyclopedia of Philosophy.

Haraway, Donna (1988). “Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective.” Feminist Studies 14: 575-599.

Kuhn, Thomas (1962). The Structure of Scientific Revolutions. The University of Chicago Press.

Laudan, Larry (1981). “A Confutation of Convergent Realism.” Philosophy of Science 49: 19-49.

Lipton, Peter (2017). “Inference to the best explanation.” In W.H. Newton-Smith (ed), A Companion to the Philosophy of Science. Blackwell.

Longino, Helen (1990). Science as Social Knowledge. Princeton University Press.

Morrow, Katie (2023). “Why we can’t resolve the scientific realism debate—and why we should believe in science anyway.” Blog of the APA.

Okasha, Samir (2016). Philosophy of Science: A Very Short Introduction, 2nd Edition. Oxford University Press.

Popper, Karl (1962). Conjectures and Refutations: The Growth of Scientific Knowledge. Basic Books.

Psillos, Stathis (1999). Scientific Realism: How Science Tracks Truth. Routledge.

Putnam, Hilary (1975). Mathematics, Matter, and Method. Cambridge University Press.

Quine, W.V.O. (1951). “Two dogmas of empiricism.” The Philosophical Review 60: 20-43.

Smith, Subrena E (2017). “Why philosophy is so important in science education.” Aeon.

Stanford, Kyle (2023). “Underdetermination of scientific theory.” In The Stanford Encyclopedia of Philosophy.

van Fraassen, Bas (1980). The Scientific Image. Oxford University Press.

Related Essays

Karl Popper and Falsificationism by Michael Zerella

Thomas Kuhn, Paradigm Shifts, and Academic Rifts by Michael Zerella

Teleological Explanations: Purposes, Functions, and Goals in Biology by Michael Zerella

Laws of Nature by Michael Zerella

Animal Minds by Tiina Carita Rosenqvist

Mindreading: Understanding Others’ Thoughts and Feelings by Emma Otterski

Philosophy of Space and Time: Are the Past and Future Real? by Dan Peterson

Philosophy of Space and Time: What is Space? by Dan Peterson

Quantum Mechanics and Philosophy I: The Superposition of Paths by Thomas Metcalf

What is Philosophy? by Thomas Metcalf

Acknowledgement

Many thanks to Jacqueline Wallis, John Kulvicki, Peter Lewis, and the editors of 1000-Word Philosophy for their helpful comments and suggestions. Thanks also to the students in my Philosophy of Science class at Dartmouth College, who enthusiastically and thoughtfully engaged with these topics during Summer 2026.

About the Author

Tiina Carita Rosenqvist is an assistant professor of philosophy at Rowan University, New Jersey. She specializes in philosophy of psychology and philosophy of medicine, with particular interests in color vision, pain, and chronic illness. TiinaRosenqvist.com

Follow 1000-Word Philosophy on Facebook, Bluesky, Instagram, Twitter / X, Threads, TikTok, and LinkedIn, and subscribe to receive email notifications of new essays at 1000WordPhilosophy.com.