Date of Award

Summer 8-15-2021

Author's School

Graduate School of Arts and Sciences

Author's Department

Philosophy/Neuroscience, and Psychology

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Positron Emission Tomography (PET) is a groundbreaking detection system that, among many other applications, enabled neuroscience researchers to detect and image physiological changes in the brain associated with cognitive operations as subjects perform a task. This technology helped researchers learn to explore the physiological basis of human cognition in new ways and was the immediate forerunner of functional Magnetic Resonance Imaging (fMRI), a technology that continues to dominate the neurosciences. PET was also the founding invention of a new field of cognitive science: human functional neuroimaging. Indeed, some would say that PET encouraged a particular way of understanding the brain – through identifying local realizations of cognition and that the ensuing field, neuroimaging, was the key experimental advance in giving rise to cognitive neuroscience properly so-called.

I use PET as a case study of technology-driven scientific change, a form of scientific change exceptionally common in contemporary science and worthy of its analysis, independently of theory- driven scientific change. I am primarily concerned with two questions, one historical and one epistemic:

• What of a general sort can the history of PET teach us about the process of technological change in science?• What can the history of PET teach us about the norms by which technologies are and should be evaluated?

To answer those questions, I have used historical, anthropological, and bibliometric research methods to reconstruct the biography of this apparatus. I have consulted the original archive of PET research at Washington University in St. Louis, interviewed most of the core researchers involved in the development of PET, and built a database of more than 100,000 articles on PET, fMRI, and neuroimaging that I then analyze to provide quantitative and qualitative information about the state of the field and the acceptance of the technology across time.

As a scaffolding framework for integrating these diverse sources of information, I rely on a core analogy to evolution by natural selection to model different aspects of the causal-historical process of technology-driven scientific change. This stands in contrast to those who understand artifacts and technologies as necessarily products of intentional choices and shaped by their intended pur- poses (Vermaas et al. 2007; Houkes and Vermaas 2010). These views emphasize the role of an intentional designer and lead us to view the engineering processes as based on the existence and ordering of preconceived and overarching goals. These goals, it is presumed, are broken into more manageable sub-goals, and then realized materially. While this image of technological development is useful as a rational reconstruction, it necessarily neglects the import of serendipity and blind variation, the centrality of material constraints and canalization, and the fluid nature of the purpose for which the technology will come to be adapted.

So, in addition to the intentional model, we should also consider a blinder, more material, more selectionist model – an evolutionary model. Technological development and technology-driven scientific change are heavily constrained, not only by the intentional decisions of designers, but also by material factors, choices of the designer’s descendants and colleagues, and shifting epistemic norms emerging and receding as a research program takes shape around the technology. The evolutionary metaphor encourages us to acknowledge aspects of this technological history that intention-driven, psychologistic accounts tend to ignore.

Many aspects of a selective process in the organic world have analogies in the evolution of detection technologies. For example, the evolutionary model leads us to consider whether modular structure of technologies enhances their “evolvability,” as modularity of organic structures is thought to do. Modules are structurally and functionally self-contained components with well-defined inputs and outputs to other components. Replacing them with functional equivalents does not impact the function of other components. Because most organisms are modular, they can mutate only certain modules to be more competitive in an environment, without needing to change or eliminate other modules that already perform at least tolerably well. Similarly, PET and technologies in general are modular; PET consists of separable components: a scintillator, a camera, an amplifier, a computer, etc. Because the technology has a modular organization, researchers could improve on one module at a time without disturbing the operation of the other components. In fact, because different modules of PET were developed primarily at different times, the modularity of PET also allows me to tell a chronological history of PET while focusing on one module at a time.

Another example, the evolutionary model leads us to consider how technologies may shape or rewrite the competitive landscape. Evolution by niche construction involves a creature selecting and structuring its environment in ways that increase its fitness. In the case of PET, scientists can try to select and structure an environment in which their apparatus has fertile applications. Like the niche construction process, competing epistemic norms select apparatus, but apparatus is not merely “selected on” by regulating norms. Researchers can actively pick and choose a less competitive environment and shape the relative priorities of epistemic norms governing data generation. I argue that, functioning like “selective forces,” a host of epistemic norms are critical to the development of detection apparatus. They can direct the development of detection apparatus, although their relative priority may be shaped by the availability of detection apparatus. I further argue that those epistemic norms complement and extend current philosophical discussion on norms of data generation.

I situate my epistemic discussion among contemporary scholars who have shifted attention to scientific practices, especially to data generation practices. Traditionally, when talking about scientific data, philosophers of science such has Hempel (Hempel 1945a, 1945b), Quine (Quine 1951), Carnap (Carnap 1966), Popper (Popper 2002), Goodman (Goodman 1983), and others primarily focused on the warrant relationship between evidence and theory: how evidence guides and should guide the construction, evaluation, justification, and revision of scientific theories. New generations of philosophers, interested in scientific practices, have expanded the attention to the epistemic and practical norms in data generation. Franklin (Allan Franklin 1989, p. 138), Staley (Staley 2004), Chang (Chang 2007, p. 86), Wimsatt (Wimsatt 2007, p. 44), and others for example are primarily interested in whether researchers can obtain data that accurately and reliably reflect target phenomena.

My work advances this discussion by emphasizing norms of data generation besides accuracy and reliability. These norms are distinct from and, I argue, irreducible to concerns about accuracy and reliability. I discuss different norms in each of the substantive chapters. These norms include increasing signal-to-noise ratio, increasing sensitivity, and increasing resolution, among others.

In short, in this dissertation, I use an evolutionary model to offer a causal-historical account of technology-driven scientific change as demonstrated by PET. In the evolutionary model, I especially emphasize that epistemic norms, such as increasing signal-to-noise ratio, are critical “selective forces” that guide the development of technologies. Those norms supplement and extend current philosophical discussions on the epistemic norms in scientific experiments and instrumentation and present a case that epistemic norms beyond and irreducible to accuracy and reliability play significant roles in guiding scientific development. This dissertation further opens up discussion about driving forces of technological change and the distinct technological and engineering goals, values, and methods reflected in those epistemic norms.


English (en)

Chair and Committee

Carl F. Craver

Committee Members

Talia Dan-Cohen, Ron Mallon, Allan Hazlett, Dennis Des Chene,