Chapter 3: Observation & Theory-Ladenness

The empiricist tradition, from Locke through the logical positivists, assumed that sensory experience provides a foundation of knowledge that is independent of theoretical interpretation. The positivists drew a sharp distinction between observation statements (e.g., “the pointer is at 3.7”) and theoretical statements (e.g., “the electron has a charge of 1.6 × 10⁻¹⁹coulombs”). Observation statements, they held, are incorrigible — they cannot be wrong — and they serve as the bedrock on which all scientific knowledge rests.

This view faced a fundamental problem. Carnap and the positivists struggled to draw the observation/theory distinction in a principled way. Where does observation end and theory begin? Is “the litmus paper turned red” an observation statement or a theoretical one (since “litmus paper” and “red” already presuppose chemical and optical theory)? The distinction seemed to be one of degree rather than kind.

“There is no observation without interpretation — without, that is, the employment of theoretical categories.”— Karl Popper, The Logic of Scientific Discovery (1934)

Even Popper, the positivists’ great critic, accepted that observation is theory-laden. He noted that observation statements (which he called “basic statements”) are always interpretations in the light of theories, and that their acceptance is ultimately a matter of conventional agreement within the scientific community, not of incorrigible perception.

N.R. Hanson (1924–1967) was the first philosopher to develop the theory-ladenness thesis systematically. In Patterns of Discovery (1958), Hanson argued that seeing is not a purely receptive, passive process. Rather, there is a conceptual component to all observation: what we see depends on what we know.

Hanson’s most famous example involves two astronomers, Tycho Brahe and Johannes Kepler, watching the sunrise. They have the same retinal image, but do they see the same thing? Hanson argued that they do not:

“Tycho and Kepler are to the sun what New-tonians and anti-Newtonians were to a falling apple. Is it the same thing they see? In one sense, yes: both have the same retinal image... But there is a sense in which what they see differs. Kepler sees the horizon dipping away from a fixed sun. Tycho sees the sun rising above a fixed horizon.”— N.R. Hanson, Patterns of Discovery (1958), p. 5

Hanson drew on Gestalt psychology to support his argument. Just as the famous duck-rabbit figure can be seen as either a duck or a rabbit, the same perceptual input can be organized in different ways depending on the conceptual framework of the observer. The shift between “seeing as” a duck and “seeing as” a rabbit is not a change in the perceptual data but in its interpretation.

Crucially, Hanson argued that theory-ladenness operates at the level of perceptionitself, not merely at the level of interpretation or judgment. It is not that Tycho and Kepler see the same thing and then interpret it differently. Rather, their different theoretical commitments cause them to have different perceptual experiences. Seeing is “theory-laden undertaking,” as Hanson put it.

Hanson also emphasized the role of background knowledgein observation. A trained radiologist sees a tumor in an X-ray where a novice sees only a blur. A botanist sees species and genera where a layperson sees only “plants.” An experienced bubble-chamber physicist sees particle tracks where a novice sees only random lines. In each case, theoretical knowledge shapes what is perceived.

The theory-ladenness thesis draws support from the psychology of perception, particularly from perceptual illusions that demonstrate the role of top-down processing in visual experience.

Thomas Kuhn radicalized the theory-ladenness thesis by connecting it to his account of paradigm shifts. In The Structure of Scientific Revolutions (1962), Kuhn argued that scientists working in different paradigms literally inhabit different worlds:

“After a revolution scientists are responding to a different world. In so far as their only recourse to that world is through what they see and do, we may want to say that after a revolution scientists are responding to a different world.”— Thomas Kuhn, The Structure of Scientific Revolutions (1962), p. 111

Kuhn provided several historical examples:

• •Priestley and Lavoisier: When Priestley observed the gas released by heating mercury calx, he saw “dephlogisticated air” (air from which phlogiston had been removed). When Lavoisier observed the same experiment, he saw “oxygen” (a distinct chemical element). Same experiment, different perceptions.
• •Aristotelians and Galileans: Where Aristotelians saw a heavy body struggling against the constraint of a string (a pendulum), Galileo saw a body that almost repeated the same motion over and over (simple harmonic motion). The Aristotelian saw constrained fall; Galileo saw a pendulum.
• •X-ray discovery: Before Röntgen’s discovery, several other physicists had noticed unusual fogging of photographic plates near cathode-ray tubes. They did not “see” X-rays because they had no conceptual framework for recognizing what they were observing.

Kuhn’s claims are often interpreted in two ways:

The strong thesis threatens the possibility of rational theory choice, because if observation is paradigm-dependent, then it cannot serve as a neutral arbiter between competing paradigms. This connects directly to Kuhn’s controversial claim about the incommensurability of paradigms (discussed in Chapter 11).

Paul Churchland extended the theory-ladenness thesis using insights from neuroscience and connectionism. In “Perceptual Plasticity and Theoretical Neutrality” (1988), Churchland argued that perception is cognitively penetrable — that higher-level cognitive states (beliefs, expectations, desires) can directly influence perceptual processing.

Churchland’s argument draws on the neuroscience of visual processing. He pointed out that the visual cortex has massive feedback connections from higher cortical areas. The information flow in the visual system is not purely bottom-up (from retina to cortex) but involves extensive top-down modulation. What we “see” is the product of a complex interaction between sensory input and stored knowledge.

“One’s perceptual experience is a function not just of the sensory input, but also of one’s antecedent theoretical convictions.”— Paul Churchland, “Perceptual Plasticity and Theoretical Neutrality” (1988)

Churchland offered several examples of cognitive penetration:

• •Expert perception: A trained musician hears harmonic relationships in a chord that a novice does not. A skilled chess player perceives strategic patterns on the board that a beginner misses. These are not merely differences in interpretation but in the perceptual experience itself.
• •Color perception and language: Research suggests that linguistic categories influence color perception. Speakers of languages with different color terms perceive the color spectrum differently, not just label it differently (the “Sapir-Whorf effect” in perception).
• •Emotional influence on perception:Studies show that emotional states affect basic perceptual judgments. Hills appear steeper when one is fatigued; ambiguous figures are interpreted in threat-consistent ways when one is anxious.

Churchland concluded that the positivist ideal of theory-neutral observation is not just philosophically problematic but neuroscientifically untenable. Perception is always shaped by prior learning, and there is no sharp line between “observation” and “theory.”

Jerry Fodor (1935–2017) mounted the most influential defense of observational autonomy. In “Observation Reconsidered” (1984) and The Modularity of Mind (1983), Fodor argued that the theory-ladenness thesis, while containing a grain of truth, is significantly overstated.

Fodor’s key move was to distinguish between the perceptual system and the central cognitive system. The perceptual system, Fodor argued, is modular: it is an informationally encapsulated input system that operates independently of central beliefs and desires. What we see is determined by the perceptual modules, not by our theories.

“The moral so far, then, is that the New Look claim that perception is importantly affected by background beliefs does not hold up well under close scrutiny... Perception may be smart, but it isn’t that smart.”— Jerry Fodor, “Observation Reconsidered” (1984)

Fodor’s central argument was based on the persistence of perceptual illusions. The Müller-Lyer illusion persists even after one knows that the lines are the same length. If perception were truly theory-laden — if beliefs could penetrate perceptual processing — then knowing the truth should extinguish the illusion. The fact that it does not suggests that the perceptual system is insulated from central cognition.

The most important property for the theory-ladenness debate is informational encapsulation: the perceptual modules do not have access to the full range of the organism’s beliefs. They process sensory input using built-in algorithms, and their outputs are then available to the central cognitive system for interpretation and judgment.

Responses to Fodor

Churchland responded that Fodor’s modularity thesis is empirically questionable. Recent neuroscience reveals extensive top-down connections in the visual system that are hard to reconcile with strict encapsulation. Moreover, Churchland argued, even if early perceptual processing is modular, the observation reports that matter for science go far beyond the outputs of perceptual modules. When a physicist “observes” an omega-minus particle in a bubble chamber, the relevant observation is heavily theory-laden, regardless of what the visual modules deliver.

Zenon Pylyshyn offered a more nuanced position, distinguishing between “early vision” (which is largely cognitively impenetrable) and “late vision” (which involves categorization and recognition and is cognitively penetrable). This suggests that the truth lies between Fodor and Churchland: basic perceptual processing may be relatively autonomous, but the observation judgments that matter for science are deeply influenced by theory.

What are the consequences of theory-ladenness for the objectivity and rationality of science? Several positions have been defended:

“Objectivity is a characteristic of a community’s practice of science rather than of an individual’s.”— Helen Longino, Science as Social Knowledge (1990)

How does the theory-ladenness thesis bear on actual scientific practice? Consider some examples from the history of science:

• •
  Eddington’s eclipse observations (1919):

  Arthur Eddington’s photographs of the 1919 solar eclipse were taken as confirmation of Einstein’s general relativity. But the data were noisy, and Eddington had to discard some plates as unreliable. Critics (including Earman and Glymour) have argued that Eddington’s prior commitment to Einstein’s theory influenced which data he accepted and which he rejected.

• •
  N-rays (1903):

  René Blondlot, a distinguished French physicist, claimed to have discovered “N-rays” — a new form of radiation. Dozens of scientists confirmed his observations. But when Robert Wood surreptitiously removed the crucial prism from Blondlot’s apparatus, Blondlot continued to “see” N-rays. The entire phenomenon was an artifact of expectation-driven perception.

• •
  Mars canals (1877–1909):

  Percival Lowell and other astronomers reported seeing an elaborate network of canals on Mars — evidence, they believed, of an advanced civilization. Better telescopes eventually revealed that the “canals” were optical illusions produced by the tendency of the visual system to connect discrete features into lines. Expectation and theory literally shaped what they saw.

These cases illustrate both the power and the limits of the theory-ladenness thesis. Theory can cause scientists to “see” things that are not there (N-rays, Mars canals), and it can influence the interpretation of ambiguous data (Eddington). But science also has mechanisms for correcting such errors: replication, independent testing, and the critical scrutiny of the scientific community.