1 The short tale of a long journey

‘Realism is dead’, intimated Arthur Fine back in 1984. Its death was announced by ‘the neopositivists, who realized that they could accept all the results of science, including all the members of the scientific zoo, and still declare that the questions raised by the existence claims of realism were mere pseudo-questions’ (Fine 1984/1991, p. 261). I share some of Fine’s sentiment.¹ But I ultimately disagree with him about the demise of realism.

My original motivations for writing this book were fairly simple and, in a way, pre-philosophical. I have always been of the view that a realist stance on science offered a safeguard to a society where trust in science was being eroded before our eyes. I watched with apprehension TV news about measles and COVID-19 outbreaks due to anti-vaccine movements gaining traction among the public; international talks on climate change breaking down under the pressure of powerful political lobbies; and scientists forced to take to the streets and march for science.

This is in no way to suggest that philosophical anti-realism underwrites an anti-science stance. Scientific anti-realism has never been a bedfellow of anti-vaccine movements or climate change denials. On the contrary, it has encouraged an appropriately critical stance towards science as a way of reflecting on its empirical foundations and resisting metaphysical grand vistas. I have always found empiricists’ arguments irresistible. And I learned from philosophical anti-realists like Bas van Fraassen (1980) never to underestimate the empirical origins of our scientific knowledge, its situated nature, or particular point of view. The story I am going to tell is thoroughly empiricist and perspectival from beginning to end.

Thus, my attraction to realism has never been a ‘call to arms’, a defence of Realism with a capital R or Truth with a capital T (I find them almost empty notions.) Indeed, ever since I was a graduate student, I have had mixed feelings about realism. Like Arthur Fine, I too was concerned that the debate on realism/anti-realism in science had obsessed over questions about ‘existence’. Do electrons exist? Do DNA strands exist? Does caloric exist? What about more complex or elusive entities such as dyslexia or dark matter?

I see now that this might not be the right question. Or necessarily the defining question about what it’s like to be a realist about science. (Hats off to the neopositivists for their foreknowledge.) Realism often felt to me like a caricature for some non-better-qualified claim about what exists. But, surely (I thought), a realist about science cannot just be someone who believes in the existence of X, Y, and Z but rejects the existence of Q, P, and R.

The recent trend in philosophy of science to go local about realism has been a salutary response to this. Realism does not have to be all-or-nothing. Realist commitments may come in degrees and shades, depending on the context of inquiry and the entities at stake (on this see Hoefer and Martí 2020). Of course, we all want to know whether what exists is the electromagnetic field rather than the ether, or viruses as opposed to miasma, or an expanding universe instead of the ancient Greek crystalline spheres.

Yet the philosophical question about realism in science does not boil down to whether X (rather than Y) exists. Questions of existence—I follow W. V. O. Quine—are best left to scientists. Realism does not have to be some kind of philosophical running commentary on what scientists have discovered. I am not suggesting that getting clear on the metaphysics of science serves no cause. It has served the cause of realism well over decades. But to my mind it has done so at the cost of neglecting other important questions.

I’d like to think of realism as an answer to first and foremost an epistemological question. To be a realist about science is to be able to answer the question of how historically and culturally situated epistemic communities have over time come to reliably advance claims of scientific knowledge.² The explanandum for realism is not what there is but how human beings come to reliably know the natural world. If one reorients the realist commitment to science along these lines—from within scientific history, rather than from a scientific view from nowhere—not only is realism alive and well: a different flavour of realism about science is within our grasp.

My starting point is, then, radically different from the concerns that over the past thirty years or so have shaped the debate on realism and anti-realism in science. The latter has revolved around the issue of whether or not our best theories in mature science tell us the truth about what there is in nature.

I won’t have anything to say about scientific theories and their being approximately true or their terms being referential. Often in this debate scientific theories were portrayed as if they had a life of their own, living in Popper’s (1978) ‘World Three’ of abstract objects. And questions of truth and reference were asked of scientific theories as we may ask of Beethoven’s Sixth Symphony whether it is ‘true’ of bucolic life, or whether the fourth movement in F minor ‘refers to’ a thunderstorm fast approaching.

My focus will be instead on the reliability of the scientific knowledge claims. That reliability matters in science is not news. Think of Philip Kitcher’s (1993, 2001) so-called Galilean strategy in arguing for the reliability of the telescope to observe ships coming to the Venetian shores no less than the craters of the Moon. Or of Naomi Oreskes’s (2019) observation that the reliability of scientific claims and the ability of scientific communities to determine it are key to the trustworthiness of science.

What is different in my treatment is the role played by scientific perspectives in assessing reliability. I see reliability not as the sort of thing that individual epistemic communities can sanction or ratify on their own. My case for perspectival realism rests ultimately on the ability to assess reliable scientific knowledge claims across a plurality of scientific perspectives. These perspectives have often spanned long stretches of time and have historically ‘interlaced’. The realism I articulate in this book arises from the deeply social and cooperative nature of scientific inquiry, with perspectival pluralism as its driving force.

The term ‘scientific perspective’ has become common in philosophy of science with Giere’s influential book (2006a) and a flurry of recent work on the topic of scientific perspectivism (see, e.g., the collection of essays in Massimi and McCoy 2019). No single generally agreed upon definition has emerged. The term lends itself to a variety of uses (metaphorical or not). For this book, I adopt the following working definition (expanding on Massimi 2018f and 2019a):

Scientific perspective (sp): A scientific perspective sp is the actual—historically and culturally situated—scientific practice of a real scientific community at a given historical time. Scientific practice should here be understood to include: (i) the body of scientific knowledge claims³ advanced; (ii) the experimental, theoretical, and technological resources available to reliably make those scientific knowledge claims; and (iii) second-order (methodological-epistemic) principles that can justify the reliability of the scientific knowledge claims so advanced.

Metaphysical, philosophical, or religious beliefs may also have been influential in making the community endorse some claims of knowledge but do not count as part of a ‘scientific perspective’ as I am going to use the term. They are intended to explain how communities come to accept some knowledge claims but not how the community comes to reliably make, or justify the reliable procedures for advancing them.⁴

A few aspects in this working definition are worth highlighting. This notion of scientific perspective extends well beyond a class of scientific models (e.g. what might be called the Newtonian perspective, the Maxwellian perspective, etc.—for this terminology, see Giere 2006a). It encompasses the whole body of claims of knowledge advanced by specific epistemic communities at any particular historical time. This is a general enough characterization to encompass claims of knowledge generated via modelling practices such as those adopted by the Intergovernmental Panel on Climate Change Assessment Report 5 that I discuss in Chapter 4.b. But it is also meant to include a much larger body of claims elicited by other experimental and technological resources. I will discuss some examples: from synthesizing hachimoji DNA in Chapter 7 to hydraulic techniques of the Alhambra engineers in Chapter 9; from local knowledge of honey-producing flora in the Yucatán peninsula in Chapter 8 to knowledge claims about Earth’s magnetic field emerging from the use of Chinese ‘wet’ and ‘dry’ compasses and their use in medieval navigational practices in Chapter 11.

From an epistemic point of view, the situatedness of scientific knowledge runs deeper than just endorsing whichever fashionable scientific theory at any particular time. Scientific perspectives, as I’d like to think of them, are effectively proxies for scientific practices, broadly understood along the lines of (i)–(iii). For it is impossible to detach the body of scientific knowledge claims from the varieties of experimental, technological, and theoretical procedures employed in advancing them reliably; and from the methodological and epistemic principles that can in turn justify those reliable procedures.

It should be clear from these remarks that my definition of ‘scientific perspective’ owes a great deal to perspectival knowledge as described by Ernest Sosa in epistemology, who charts a middle ground beyond foundationalism and coherentism.⁵ Sosa draws an important distinction between what he calls apt beliefs and justified beliefs. The former are reliably obtained at a first-order ‘animal knowledge’ level that we share with non-human animals. For example, my reliable belief that milk is in the fridge is something I share with my cat. Justified beliefs, by contrast, belong to a second-order ‘reflective knowledge’ level.⁶ Here a perspectival ascent to an epistemic perspective is required in order to reflect on the sources of the reliably forming methods and procedures behind apt beliefs.

The epistemic perspective includes, then, first-order reliable claims of knowledge about the objects under investigation and second-order methodological and epistemic principles that justify the reliability of the experimental, theoretical, and technological resources used to make the first-order claims. There are a number of attractive features in this way of thinking. The first is that a clear distinction between truth and justification for claims of knowledge becomes immediately available. The truth of knowledge claims endorsed by particular epistemic communities is ultimately a matter of correspondence with the way the world is and depends on having reliable experimental, technological, and theoretical procedures for arriving at these claims. How those reliably formed claims are in turn justified is, however, perspectival.⁷

What changes when historically shifting from one scientific perspective to another are not the reliably formed claims of knowledge (if they are indeed reliably formed), but the epistemic-methodological justificatory principles. The reliability and ultimately the truth of those knowledge claims is not fixed by the scientific perspective in which they might have originated. Scientific perspectives do not offer perspectival facts. Nor should truth be understood in terms of perspectival truthmakers (as I point out in Chapter 3), or as indexed to a perspective or relative to a perspective. As I explain in Chapter 5 (building on Massimi 2018e), while I see scientific perspectives as offering both justificatory principles and assertability conditions for specific claims of knowledge, I do not see truth conditions as something to be delegated to any specific scientific perspective.

As new scientific perspectives come to the fore, existing scientific knowledge claims can be cross-perspectivally assessed and retained or withdrawn accordingly. While truth as correspondence with the way the world is (loosely speaking) a cross-perspectival affair, scientific perspectives offer a second-order set of epistemic-methodological principles that can shed light on whether or not someone has justifiably come to reliably formed claims of knowledge. Therefore I see the plurality of scientific perspectives not as a disjoint set but as intersecting with one another to fulfil this crucial epistemic role for scientific knowledge.

It could be, for example, that while reliably formed, some scientific knowledge claims are suffering from justificatory principles that might be defective, or insufficient by themselves to ground the reliability of the procedure.⁸Cross-perspectival assessment of scientific knowledge claims is key to deliver scientific knowledge. Hence the heavy lifting done by the pluralistic, diverse, and fluid interplay of scientific perspectives.

I come back to the distinctively philosophical question at the heart of perspectival realism: how do we human beings come to know the natural world as being a certain way from a number of historically and culturally situated perspectives? When posed this way, the philosophical question about realism is less about mapping the existence of the ‘scientific zoo’, and more about exploring what makes us wonderfully diverse human beings capable of reliable scientific knowledge over time. Perspectival realism, as I articulate it in this book, is not a metaphysics-first view. It is a project in the epistemology of science.

There are two main and interrelated motivations behind perspectival realism as a project in the epistemology of science. The first is historical. I have always thought that epistemic stances about science (realism, instrumentalism, empiricism, or similar) should be able to speak to the history of science. This is not a rehash of ‘philosophy of science without history of science is empty’. It is more the need to take seriously the historically situated nature of scientific knowledge when discussing its epistemic foundations. My overarching question demands engaging with the historical plurality of practices in the sciences.

Understanding how we have come to know the world as teeming with atomic nuclei, melliferous flowers, DNA strands, and so forth, requires understanding how particular epistemic communities at particular historical times have produced reliable knowledge claims about them. Moreover—and the most important part of my view—it requires understanding why the associated realist commitments could be retained across different epistemic communities over time (or cross-perspectivally, as I call it).

The second and related motivation has to do with what I am going to refer to loosely as ‘multiculturalism’ (knowing all too well that the term has acquired a very specific meaning in political theory which is not necessarily how I am going to use the term here). Debates on realism in science and scientific knowledge more broadly have a tendency to proceed in some weirdly engineered cultural vacuum. There are obviously methodological reasons why philosophers of science cash out narratives in terms of ‘claims of knowledge’, ‘inferences’, ‘suppositional antecedents’, ‘indicative conditionals’, and so forth. These are some of the epistemic-semantic tools in our profession, as much as quadrupole moments, spin, and so on, are tools for nuclear physicists. But there is a problem arising from the uncritical use of such philosophical tools.

They hide a presumption that scientific knowledge production proceeds on some kind of idealized frictionless plane rather than in well-defined historical and cultural contexts that affect the nature of the claims of knowledge advanced. One of the motivations for perspectival realism is to counteract this presumption. The realism I shall be articulating is realism within the bounds of a plurality of intersecting scientific perspectives. Therefore, I understand the notion of scientific perspective rather broadly to include any scientific practice that has resulted in reliable knowledge claims that have been cross-perspectivally retained. This implies re-assessing the role played by a great number of historically and culturally situated epistemic communities in knowledge production, especially those communities that are often severed by epistemological narratives and frictionless accounts of scientific knowledge. I am thinking, for example, of the local knowledge about the melliferous flora among the beekeepers of the Yucatán peninsula (Chapter 8); or about the rosy periwinkle in the communities of South Madagascar (Chapter 11); or about ‘kelp-making’ (i.e. producing ashes of seaweed used in glass production) by the Scottish Hebridean communities of the eighteenth and early nineteenth century (see Chapter 10). Similarly, my working notion of scientific perspective extends to the engineering practices of studying ground-water motion to build magnificent fountains in parks and public spaces, such as the fountains of Alhambra and Villa d’Este (which I discuss in Chapter 9).

In Chapter 11, I clarify the multicultural dimension in the notion of ‘scientific perspectives’ and the implications for how to think of scientific knowledge production and associated epistemic injustices. In particular, I urge us not to think of scientific perspectives in terms of ‘shared membership’, or ‘shared scientific homeland’, or disjoint scientific ‘silos’, but instead as historically ‘interlacing’ and stretching beyond specific geopolitical and national boundaries. While ‘intersecting’ is a methodological feature of how scientific perspectives can be brought to bear on one another to refine the reliability of particular claims of knowledge, ‘interlacing’ is a historical feature. It refers to how situated perspectives have encountered and traded with one another some of their tools, instruments and techniques over time. ‘Interlaced’ perspectives track the evolution of knowledge concerning particular phenomena in what I call a ‘historical lineage’.

When seen through the lenses of perspectival realism, scientific knowledge becomes knowledge whose reliable production is not the prerogative of one single community at one historical moment. It is social and collective in a distinctively multicultural way. This view has far-reaching implications for two historical kinds of epistemic injustices about scientific knowledge: what in Chapter 11I call epistemic severing and epistemic trademarking. Epistemic severing is the almost surgical excision of the contribution of particular communities (either within the same scientific perspective or across culturally diverse perspectives) from narratives about scientific knowledge production. Epistemic trademarking is the subsequent fencing and ultimately often merchandising of portions of scientific knowledge as a ‘trademark’ of one epistemic community at the expense of others who have historically contributed to such production.

I see perspectival realism as an antidote to these epistemic injustices and a prelude to what at the very end of the book I call non-classist scientific cosmopolitanism, taking my cue from a large literature in cultural studies, sociology, and anthropology: scientific knowledge at the genuine service of a diverse and multicultural ‘world citizenship’.

The book is in two parts. In Part I (through Chapter 5), I delve into the epistemology of science. I analyse scientific practices which I cluster under the heading of ‘Perspectival Modelling’. In Part II (Chapters 6 through 11), entitled ‘The World as We Perspectivally Model It’, I clarify the kind of realism that emerges. Perspectival realism comes down to what the perspectivally modelled world is going to look like.

I see perspectival modelling as a specific variety of model pluralism in science. The plurality of models here is ‘exploratory’ in enabling a particular kind of inferential reasoning that proves fruitful when we want to explore what is possible. This exploration, I argue, is an important guide to what is actual in science. Indeed, it is often the only way to find out what is actual given that we do not have a ‘view from nowhere’ on nature.

In this respect, perspectival models, as I understand them and use the term in this book, are not necessarily autonomous entities mediating between the theory and the experimental data (along the lines of the influential view of models as mediators articulated by Mary Morgan and Margaret Morrison [1999]). Perspectival models are not necessarily downstream from higher-level theories, even if theories are always at work in the modelling. Nor is their role to map onto or be isomorphic to particular empirical data, or patterns thereof.

Instead of taking scientific theories in mature science as my starting point, I take perspectival modelling—with its exploratory role—as the very starting point of scientific inquiry. I understand perspectival modelling as an integral part of scientific perspective in the broad sense in which I have defined it in Section 1.2. It is important to distinguish ‘perspectival modelling’ from ‘perspectival models’ as I do in Chapter 4.

The latter are a variety of scientific models. Perspectival modelling (as I use the term) is not restricted to scientific models exclusively. It refers more generally to the situated modelling practices of epistemic communities, including the way they use particular experimental and technological resources to advance claims of knowledge and make inferences from data to the phenomena of interest. It captures how situated epistemic communities across a number of intersecting scientific perspectives come to know the world as being a certain way, as I illustrate with my three case studies in Ch. 4.a, 4.b, and 4.c. Perspectival modelling is a general expression under which I include the situated practice of dendroclimatologists no less than that of CMIP5 climate modellers; the respective practices of educationalists and of developmental psychologists working with models for dyslexia; the practices of petrologists, cosmochemists, and physicists working with nuclear models. This is the distinctive way in which I use the term ‘perspectival modelling’ in Part I of this book.

Given this terminological distinction between ‘perspectival modelling’ and ‘perspectival models’, one should tread carefully here. What makes some modelling ‘perspectival’ is not that each model involved in the wider modelling practice offers a different perspective on a given target system (as one might be tempted to think of it in a colloquial sense). What makes some modelling ‘perspectival’ is instead its being embedded into historically and culturally situated scientific practices of particular communities and its fulfilling a distinctively exploratory role in delivering scientific knowledge over time.

Zooming into ‘perspectival models’, I characterise them as inferential blueprints: they enable a variety of situated epistemic communities over time to make inferences from a range of datasets to what I call modally robust phenomena. Thus, I see perspectival models as representing, and in turn, I understand their representational function in a broadly inferentialist way (following on Suárez’s very influential view—see Suárez 2004, 2009, 2015a, 2015b).

The book begins with a discussion of how to understand the notion of perspectival representation, a theme that has intrigued me since the time of Bas van Fraassen’s (2008) book. The metaphor of perspective has featured in the works of philosophers of science—from Ron Giere to Paul Teller, from Bas van Fraassen to Sandra Mitchell, among many others—who have appealed to the importance of perspectivism. Yet the metaphor must be handled with care.⁹ What makes our representations—in science as well in art—perspectival in some philosophically interesting sense? In Chapter 2, I draw attention to two (complementary) ways in which a representation (in art or in science) can be said to be ‘perspectival’.

Think of your favourite perspectival drawing in art. What do you see? You are likely to see the scene being represented from a particular angle. If the painting concerns an interior, you might see the objects lined up, with the ones in the foreground being bigger than those in the background. If it is a landscape, trees and figures are displayed along coplanar lines, giving the impression that you are observing the scene from below, or from above; from the centre or from left or right. You might even see the landscape opening up in different directions if more than one vanishing point is used. In any case, what you are likely to see is something that looks very much like a ‘window on reality’, a space where objects are situated, rather than a chaotic aggregate of objects of all sizes piled up on each other. The represented scene follows coplanar lines towards vanishing points and situates the objects along it, the bigger in the foreground and the smaller in the background, creating a sense of space and depth in what is effectively only a two-dimensional canvas.

Now think of what it means for a representation to be ‘perspectival’. You might reply that a representation can be perspectival₁ because it is drawn from a particular vantage point: the interior or the landscape is represented as drawn from a particular angle where you can see some objects more prominently than others. But a representation can also be said to be perspectival₂ because it is drawn towards one or more vanishing points. These two ways of thinking about what makes a representation perspectival are two sides of the same coin. It is because the representation has one or more vanishing points that it appears to be drawn from a particular point of view.

There does not have to be a tension between these two ways of thinking about what makes a representation perspectival. They are clearly compatible: any representation can be said to be perspectival both because it is drawn from a particular vantage point and because it is directed towards one or more vanishing points. Yet the emphasis that we place on one of these two ways of thinking has philosophically far-reaching consequences. The first notion stresses the situatedness of the representation. The second its directionality.

In Chapter 3, I disentangle a host of issues that are often found in association with the first notion in the literature on scientific modelling and scientific representation. I return to it in Chapter 11, where I unpack what I take to be the most important insight in the idea of situated representation. How should one understand the idea of historically and culturally situated scientific perspectives? How not to conflate it with the deceptively similar idea that scientific perspectives are somehow well insulated from one another, or defined by some kind of shared membership, maybe one that is restricted to specific temporal and geopolitical boundaries? In Chapter 11, I give my reasons for seeing scientific perspectives as historically ‘interlacing’ over long periods of time. This is what makes scientific knowledge production possible at all, as knowledge that belongs to all human beings (or cosmopolitan scientific knowledge, as I call it).

Before then, Chapters 4 through 10 focus on the second notion of perspectival representation. This is closely patterned on the analogy with perspectival drawings in art. A representation is perspectival₂ in having one or more vanishing points that transform a two-dimensional canvas into a three-dimensional ‘window on reality’, to echo art historian Erwin Panofsky (1991). Perspectival models, I contend in Chapters 3–5, offer a ‘window on reality’ because they make it possible for human beings to make model-based inferences about phenomena and ultimately natural kinds. They act as ‘inferential blueprints’. The central idea of inferential blueprints is meant to capture a set of salient features of this modelling exercise:

These inferences are often couched in terms of indicative conditionals with the following form: ‘if x is the case, y will be the case’. I see these indicative conditionals as having antecedents (if x is the case) that invite us to imagine or ‘physically conceive’ some scenarios suggested by the model. And I see their consequents (y will be the case) as hiding an epistemic modal such as ‘can’ or ‘may’ or ‘might’ (following Angelika Kratzer’s 2012 view of epistemic conditionals). In this way, I maintain, perspectival models enable inferential reasoning that over time and often through a plurality of models employed by different epistemic communities allows us to explore what is possible and gain a ‘window on reality’.

But what can one see through this ‘window’? What is the perspectivally modelled world going to look like? Part II of the book delivers my take on realism in science, a version of realism that is downstream from the epistemological framework of perspectival modelling broadly understood. I characterize the realism that emerges from perspectival modelling as ‘bottom up’: from data to phenomena to natural kinds. This is in contrast with familiar realist views that start with scientific theories, their main theoretical posits and theoretical terms, and ask which elements of reality correspond (or not) to those.¹⁰ It is also in contrast with metaphysics-first approaches that engage in reading essential properties, dispositions, categorical properties off our current best theories. Reading realism through the kaleidoscope of scientific perspectives forces upon us a shift from metaphysics-first to epistemology-first. A move away from searching for what Gilbert Ryle (1949/2000) aptly called the ‘hidden goings-on’, towards the scientific practices that are the inferential sources of modally robust phenomena.

Thus, the realism emerging from perspectival modelling is not realism about unobservable entities or similar. It is realism about phenomena that do not just occur, but could occur under a range of different experimental, theoretical and modelling circumstances and across a variety of perspectival data-to-phenomena inferences. These are not the phenomena of scientific realists or anti-realists. They are neither faint copies of the real unobservable entities nor sheer empirical appearances. Their modal robustness is reminiscent instead of the Kantian ‘objects of experience’ as the only objects one can claim knowledge of.

Therefore my phenomena-first ontology departs from traditional empiricist views of phenomena as mere appearances of the unobservable entities beneath them. It follows instead a long-standing neo-Kantian tradition in packing enough modal strength into the notion of phenomena for them to do the heavy lifting when it comes to realism in science.

I present some of these modally robust phenomena in my three case studies in Part I: nuclear stability in Chapter 4.a; global warming in Chapter 4.b; and children’s difficulties with reading in Chapter 4.c. I describe how one comes to know each of these phenomena through a plurality of intersecting scientific perspectives and associated perspectival data-to-phenomena inferences. We do not encounter in our scientific travels a realm of unobservable entities. Nor do we stumble upon a sparse Humean mosaic of natural properties with no causal glue in between them. The perspectival realist landscape that opens up is not inhabited by categorical properties, dispositional essences, potencies, tropes, or universals.

It is instead populated by modally robust phenomena such as the bending of cathode rays, the decay of the Higgs boson, the electrolysis of water, germline APC mutations, the pollination of melliferous flora, the growth of a mycelium, among countless other examples.

In Chapter 6, I tease out the realist commitment to modally robust phenomena via what I call the evidential inference problem: on what distinctively epistemic grounds do data provide evidence that a particular phenomenon is real? I argue that answers to this problem point towards phenomena understood as stable events that are modally robust across a variety of perspectival data-to-phenomena inferences. I understand the stability of the event in terms of what I call ‘lawlike dependencies’ among relevant features. And I see modal robustness as a secondary quality that has to do with how a plurality of historically and culturally situated communities are able to tease out the network of inferences from a variety of datasets to the stable event in question. Stable events, in other words, act as the realist tether in guiding epistemic communities through a maze of alternatives in being able to identify and re-identify over time the phenomena in question as modally robust.

But obviously, it is not enough to say that perspectival realism is realism about modally robust phenomena. Traditionally, realism in science is associated with natural kinds. And realism about kinds has often come with a defence of metaphysical posits, be they categorical properties or dispositional essences. Scientific realists have often argued that a commitment to these metaphysical posits is compatible with scientific pluralism, maybe in the way in which these properties are clustered together (think of Anjan Chakravartty’s [2007] so-called sociable properties coupled with dispositionalism). In Chapters 7 to 10, I delve into the topic of natural kinds and offer a view which I call ‘Natural Kinds with a Human Face’ (NKHF). In brief, I define natural kinds as the

A few aspects are worth highlighting in this definition. First, not any grouping of phenomena counts as a NKHF. In Chapter 7, I distinguish among what I call empty kinds (e.g. caloric), in-the-making kinds (e.g. dark matter), and evolving kinds (e.g. electron) depending on their respective ‘nomological resilience’ across a number of perspectival data-to-phenomena inferences. Laws of nature, or, better, lawlike dependencies, are key in transforming over time any in-the-making kind into an evolving kind. As they are also key in revealing if an in-the-making kind may in fact be an empty kind (in the absence of suitable lawlike dependencies). I argue that all natural kinds we know and love are ultimately (and fallibly) evolving kinds as the groupings are always open-ended. Lawlikeness, under this view, is not supervenient on dispositional essences or categorical properties but is inherently a primitive relation among features of events that are candidates for modally robust phenomena (as I spell out in Chapters 5 and 6).

The other aspect worth highlighting in NKHF is its underlying historical naturalism. In Chapter 8, I argue that the naturalness of natural kinds is not necessarily the expression of natural joints through which nature comes pre-carved. It is rooted instead in stable events and the lawlike dependencies inherent in them. But it also depends on our perspectival scientific history—how historically situated communities have come to identify and group relevant phenomena together over time. Historical naturalism as such is the first step in my broadly inferentialist and anti-foundational view of natural kinds, which takes inspiration from Neurath’s Boat. It is also a necessary component for both the historical and multicultural motivations I mentioned earlier. Shifting attention from ‘scientific theories in mature science’ to ‘modally robust phenomena’ (and how to perspectivally find them) can realign the debate on realism away from Western-centric narratives.

Since the notion of scientific perspective pays attention to the modelling, experimental and technological resources available to any epistemic community to reliably advance scientific knowledge claims, the ability to identify and re-identify modally robust phenomena becomes a collective enterprise. One to which, for example, ethnobotany contributes alongside cytogenetics, when it comes to botanic taxa, as I discuss in Chapter 8. Or think of how medieval hydraulic engineering in Alhambra was instrumental to gaining knowledge of the phenomenon of ground-water motion, which alongside other phenomena (from water droplet formation to chemical bonds) is associated with the natural kind ‘water’. In a genuinely anti-foundational Neurathian spirit, each NKHF is an open-ended grouping of historically identified phenomena that different situated epistemic communities have robustly identified over time. NKHF do not take any particular phenomenon (and its associated lawlikeness) as more essential or foundational than any other.

However, NKHF are not another name for conventionalism about kinds. In Chapter 9, I clarify the specific brand of contingentism at play in the view and why it is not to be confused with conventionalism. I will cash out the Spinoza-inspired view of natural kinds as sortal concepts and show that what holds open-ended groupings of phenomena together is a sort-relative sameness relation.

The last element of my account of NKHF has to do with their inferential nature and in particular what I call truth-conducive conditionals-supporting inferences. In Chapter 5, I discuss the semantic nature of these inferences, and in particular the role of indicative and subjunctive conditionals. In Chapter 10, I develop this aspect by delving into how scientists between 1897 and 1906 came to identify the electric charge as a fundamental unit and property of the natural kind ‘electron’ qua an evolving kind. Once again, I illustrate the role played by a number of historically and culturally situated scientific perspectives in knowledge production. Some of these were historically ‘minority scientific perspectives’, perspectives within Western science that have been forgotten, or erased in the grand historical narrative of scientific realism.

I share the sentiments of many colleagues in history of science in thinking that such minority perspectives do not belong to the dustbin of history. Our philosophical tendency to regard them as intellectual curiosities usable at best for case studies shows how pervasive the narrative of winners vs. losers has been in this debate. My goal is to show that they played a vital part in enabling the chain of inferences that eventually led epistemic communities to reliably advance scientific knowledge claims over time. They were an integral part of how culturally situated scientific knowers reliably and justifiably have come to know the natural world as being a certain way.

We remember today J. J. Thomson as the discoverer of the electron. But he called it a ‘corpuscle’. He believed corpuscles to be end points of what he called a ‘Faraday tube’ as a field-theoretical fin de siècle scientific perspective on the nature of electric charge. This was soon to be overtaken by the scientific perspective coming out of Max Planck’s work on black body radiation. We have similarly forgotten the complex experimental practices that made Thomson’s work on cathode rays possible in the first instance. These included mastering the glass-blowing techniques for making bespoke cathode ray tubes for his researches. But also practices such as ‘kelp-making’ (collecting ashes of burnt seaweeds) among local communities of the Scottish Islands (e.g. the Hebrides, Orkney, among others), which enabled the British glassware industry to flourish at the start of the nineteenth century and produce the lead-free glass required for electrical researches.

Along the same lines, the short-lived engineering practice of the ice calorimeter devised by Lavoisier and Laplace at the end of the eighteenth century proved important for identifying the relevant phenomena that eventually led to transitions of states (from solid ice to liquid water) to be regarded as physical (rather than chemical) in nature, as I discuss in Chapter 7. These examples show how modelling and experimental practices that were an integral part of what turned out to be two ‘minority perspectives’ from a historical point of view played in fact a pivotal role for knowledge production at two key historical junctures. They also show the crucial role that artisans, glass-blowers, hydraulic engineers, beekeepers, and kelp-makers, among others, play in making reliable scientific knowledge possible within and across situated scientific perspectives.

I return to this aspect at the end of the book, in Chapter 11, where I revisit the situatedness of knowledge, and in particular the first notion of perspectival representation introduced in Chapter 2. Scientific perspectives offer standpoints afforded by particular geographical, socioeconomic, political, historical, and cultural locations. Yet collectively over time they intersect with one another and make scientific knowledge production possible. This becomes an opportunity for briefly engaging with the important topic of multiculturalism and cosmopolitanism in science. The way in which situated scientific perspectives span over time and historically ‘interlaced’ deserves more attention to tease out when this has been virtuous (as opposed to exploitative) in producing knowledge. I give the broad normative contours of this process in Chapter 11, where I discuss two varieties of epistemic injustice I have mentioned already: epistemic severing and epistemic trademarking. I argue that they demand more than mere ‘recognition remedies’. They call for the ‘reinstatement’ of epistemic communities that have been unjustly excised from narratives of scientific knowledge production as a stepping-stone to a non-classist and non-elitist variety of what I call ‘cosmopolitan science’.

Perspectival realism tells a story about realism in science with no winners or losers, no dominant ‘paradigm’ taking over an old one ‘in crisis’, or ‘progressive research programme’ gaining ground over a ‘degenerating’ one. The kind of realism that I offer in this book does not treat the history of science as the base for writing the epistemic hagiography of the winners. It embraces pluralism in methodologies and models, and a commitment to a plurality of situated scientific practices. It sees the many varieties of local knowledge as an integral part of scientific knowledge production. It takes our engineering practices and synthetic kinds as continuous with our natural kinds.

I agree with van Fraassen that on matters of realism, instrumentalism, and so on, there are ultimately ‘voluntaristic stances’. Thus, I do not harbour hopes to convert the die-hard metaphysicians to swap their ontology of dispositional essences or categorical properties for a phenomena-first ontology. I will rest content if I manage to put forward—as carefully as I can and with as many examples as I can muster—an alternative way of thinking about realism, a different lens through which one can see the ways in which realism and perspectival pluralism in science could fruitfully come together. I ultimately want to celebrate scientific knowledge as a distinctive kind of social and cooperative knowledge—knowledge that pertains to us wonderfully diverse human beings occupying a kaleidoscope of historically and culturally situated perspectives.

If ever there was anything miraculous about the success of science, it is not the success of an individual scientific theory T in latching onto a mind-independent world. It is the success of our human species in building over time an extraordinarily varied and reliable knowledge of the world we live in, despite having no privileged standpoint to occupy. What needs to be explained therefore is how historically and culturally diverse epistemic communities have achieved this epistemic feat. This book offers a possible story about it.

I believe there is a lot of work for philosophers of science still to do in this respect. Historically, science has too often been for the benefit of Western elites. Philosophers of science, as I see it, have a duty not to be self-complacent in writing narratives about science and scientific knowledge. How one ought to think about science is different from how science has historically happened. And paying attention to the historical practices should not become an automatic way of epistemically ratifying the underlying mechanisms of knowledge production (especially whenever exclusionary mechanisms might have been at play).

The tension between the normative and the descriptive components in how we think about science is tangible in the discipline of history and philosophy of science (HPS). But philosophers have a duty to reorient those narratives—and realism is a key one—in a way that does not end up ratifying templates of epistemic dominance and exclusion. I see our job as that of reflecting on the practice of science and laying out ways in which one ought to be thinking about scientific knowledge.

To conclude, let me return to the metaphor of perspective. The vanishing point(s) towards which the lines of perspective converge in a drawing should not be understood as a proxy for a metaphysical reality to which we all converge at the end of inquiry. After all, they are only vanishing points. Perspectival realism is not a form of convergent realism. It is not forced upon us by any metaphysics-first approach to the natural world.

The absence of a scientific ‘view from nowhere’ means there exists no ideal atlas, and no privileged catalogue of ontological units. There is no directionality to serve in worn-out discussions about convergence towards a final theory or a final metaphysical reality or end of inquiry. Nor one that is synonymous with consensus building as a way of homogenizing lines of inquiry and smoothing out pluralism. There is, in other words, no conclusion to the story of our scientific endeavours. Scientific knowledge is ongoing and open-ended.

I am often asked what perspectival realism is. It has occasionally been suggested to me that we can think of this asymptotically, in the same metaphysically innocent way in which the American pragmatist Charles Sanders Peirce envisaged the ‘end of inquiry’ as a regulative ideal rather than a factual state of affairs realized at any point in time. But my view of the directionality of perspectival₂ representations (their having one or more vanishing points) in no way resembles convergence to something, even in the mild regulative Peircean sense.

The directionality I describe is more like that of Marco Polo, leaving Venice to explore uncharted territories in the East. His journey was not guided by any universal atlas. Nor, similarly, does science provide any universal atlas to nature and all that it contains. We are all fellow travellers, together with our predecessors and successors, across countries and myriad cultures in the world.

Being a perspectival realist about science is therefore akin to walking through the garden of inferential forking paths (to echo Jorge Luis Borges’s 1941/2000 tale). I’d like to think of the directionality of perspectival representations as the directionality of fellow travellers zig-zagging through the garden, with no ideal atlas, but with plenty of inferential resources for reliably tracing and retracing our journeys, and not getting lost or permanently stuck along the way. That’s how we encounter nature as teeming with modally robust phenomena and Natural Kinds with a Human Face.

I grew up in a small village in the Sabine hills near Rome: one of those unassuming places of a few thousand inhabitants where not much goes on and the days, weeks, and months all resemble one another. The long summer holidays were the toughest to live through. The place came to a standstill from June to September in the unbearable heat. Only the relentless sounds of the cicadas kept me company during this season as a teenager. And my books, of course, mostly novels that kept me amused while the rest of the world outside seemed to fade away under the scorching temperatures.

I began to read Italo Calvino’s books. I loved his short stories and novels permeated by realism sprinkled with a good pinch of imagination. I read Invisible Cities and thought of cities I’d like to see, new places I’d like to explore as in Calvino’s book the Chinese ruler Kublai dreamt of imaginary cities he had never visited and asked the Venetian Marco Polo about them.

This book too is about journeys of exploration—though not into imaginary cities but the natural world. It is a philosophy of science book that looks at scientific knowledge as an exploratory journey and tells a story of our travels through it. A bit like in Calvino’s stories, this too is a story of realism about science combined with a good pinch of imagination. In this case, the imagination concerns how we model the way the natural world might be at every step and junction along the route and how we come to encounter a natural world teeming with phenomena and natural kinds as a result of it. In Chapter 5, I call this kind of imagination ‘physical conceivability’ in the context of a particular type of scientific modelling—perspectival modelling—to which Part I of the book is dedicated.

By contrast with Calvino’s Marco Polo, whose fervent imagination could conceive of a myriad of possible cities, our encounter with the natural world requires hard work on our part. We have to find our ways through swarms of data and empirical regularities with the help of cleverly designed modelling practices, and always within the boundaries afforded by the technological, experimental, and theoretical resources available in any culturally located perspective at any one time. I owe my other literary debt to Jorge Luis Borges’s Fictions, and in particular the aforementioned short story ‘The Garden of Forking Paths’, which provides the background reference for my inferentialist account of natural kinds from Chapter 8 onwards.

My autobiographical journey of exploration led me out of the Sabine hills of Lazio in the late 1990s, almost catapulting me into thriving London and the buzzing London School of Economics, and taking me through lands and places surpassing my wildest teenage dreams. I would not have made it thus far without the help and support of countless fellow travellers along the way—far too many to recount and thank here.

But to some I owe a special debt of gratitude for having made this book possible. Sadly, some of these people have passed away and did not get to live long enough to see the end of this particular journey. But it is to them and their enthusiastic encouragement of my work at times when I was myself in full-swing self-doubt that I owe very special thanks.

My interest into perspectivism began unexpectedly in 2007 with an email I received from Peter Lipton. I was in Cambridge at the time as a Junior Research Fellow at Girton College and I regularly attended Peter Lipton’s Epistemology Reading Group on Thursday afternoons. We read Langton’s Kantian Humility but also Putnam’s Reason, Truth and History and Nickles’s edited collection on Thomas Kuhn.

One day, Peter sent me a copy of his book review of Ron Giere’s Scientific Perspectivism, knowing about my ongoing fascination with Kant and Kuhn and Putnam’s internal realism. In that book review, he located Giere’s book in a tradition that he saw as beginning with Kant and continuing with Kuhn: a tradition he called ‘constructivism’. Taking his cue from Giere’s analysis of colours as irreducibly perspectival, Peter Lipton concluded that, ‘Like Kant, Giere wants to extend his picture of colours to all science. Scientific descriptions capture only selected aspects of reality, and those aspects are not bits of the world seen as they are in themselves, but bits of the world seen from a distinctive human perspective’ (Lipton 2007, p. 834).

Peter Lipton himself was never persuaded by what he saw as the Kantian line behind Giere’s perspectivism. His review concluded: ‘Maybe in the end constructivism is true, or as true as a constructivist can consistently allow. Nevertheless, the thought that the world has determinate objective structures is almost irresistible, and Giere has not ruled out the optimistic view that science is telling us something about them’ (Lipton 2007, p. 834). His scientific realist reservations notwithstanding, he did send me the review with the accompanying note that he thought I was going to like this book.

He was right. I did like Ron Giere’s book. I more than liked it. Before I knew it, that book set me on a path that I have been on for the past fifteen years. My dialogue with Peter Lipton continued as we were planning a joint Philosophy of Science Association 2008 symposium where he was going to give a talk aptly entitled ‘Kantian Kinds and Natural Kinds’. But Peter’s untimely death in November 2007 marked an abrupt end to our exchange.

The exchange continued indirectly when in 2010 Anjan Chakravartty edited a special issue of Studies in History and Philosophy of Science in Peter Lipton’s memory in which he himself published a seminal article on perspectivism (Chakravartty 2010). To Anjan I owe many thought-provoking conversations over the years on the topic of realism and perspectivism that often made me pause and think hard about some of the issues I cover in this book.

In 2008, Bas van Fraassen’s book Scientific Representation: Paradoxes of Perspective came out and it provided a further opportunity for me to get more familiar with the literature on perspectivism. I loved the subtle interplay between art and science in the book, between Alberti’s and Dürer’s pictorial perspective and the use of machines and engines to offer representations of phenomena not as they are—van Fraassen says with a subtle Kantian undertone—but as they appear from the particular vantage point of an observer.

I wrote a review of Bas van Fraassen’s book where I briefly touched on the Kantian theme of the distinction between phenomena and appearances (Massimi 2009, p. 326). I had been following his constructive empiricism since my graduate days. He too, like Peter Lipton, showed great generosity in replying to my philosophical questions and engaging with my often tentative (and mostly half-baked) ideas. I was thrilled to see the Kantian line becoming more tangible in his 2008 book. That was the time when I was embarking on my philosophical detour into Kant’s philosophy of nature. My interest in Kant goes back to my undergraduate times and to my then dissertation supervisor Silvano Tagliagambe at the University of Rome. His Kantian-inspired epistemology provided the fertile breeding ground for my research interests in this area in the decades that followed.

I have never thought of Kant as a ‘constructivist’, despite Peter Lipton’s assessment, unless one understands the word in a rather broad sense—in the same sense in which, for example, van Fraassen is a ‘constructivist’ in taking scientific models as human constructions that save the observable phenomena. Hence the self-declared ‘constructive empiricism’ of his view. By the same token, Ron Giere’s Scientific Perspectivism has never struck me as a ‘constructivist’ piece of work. I never asked Ron, but I suspect he would have rejected the label for himself, unless again one understands it in a broad sense. But understood thus, the term ‘constructivism’ loses most of its philosophical force. In a way, we have all been constructivists all along so long as one takes scientific models as human constructions designed to explore the natural world. (After all, what are scientific models if not human constructions?)

To put it differently, insisting on the ‘constructivist’ label (for Kant, or Kuhn, or van Fraassen, or Giere) seems to be missing a crucial point here. Namely, that scientific realism qua ‘the optimistic view that science is telling us something’ about the world’s ‘determinate objective structures’ (Lipton 2007, p. 834) cannot evade the following puzzle: how do scientific models qua human constructions match or fit or accurately represent ‘determinate objective structures’? If they are human constructions, how can scientific models ever claim to even rise to the challenge of solving this puzzle between objective reality, on the one hand, and the scientific image, on the other hand?

This epistemological question cannot be eschewed by off-loading the explanatory task to metaphysics, in my view. Or by packaging the natural world with causal properties, dispositional essences, causal powers, and so forth. The answer to the epistemological question ‘How is scientific knowledge of the natural world possible?’ cannot be outsourced and delegated to an arsenal of presumed metaphysical entities, whose very reason for existence is precisely to take off our shoulders the uncomfortable burden of the question.

This same question, of course, led Kant to give his own answer without taking refuge in any metaphysical realm. In my view, the particular answer Kant gave to it does not make him a ‘constructivist’ any more than Giere’s answer to the same question in terms of scientific perspectivism, or van Fraassen’s answer to it in terms of empirical adequacy, makes either of them ‘constructivist’ in some genuine sense.

If anything, Kant described his view as a kind of ‘empirical realism’, but of a very nuanced kind. He was someone for whom reality consisted of phenomena, but phenomena are not mere copies of things, or faint images of some underlying unobservable realm. Kant’s empirical realism—I thought—sat very well with van Fraassen’s take on scientific representation and perspectivism, much as van Fraassen would of course resist the label ‘realism’ altogether.

As the years have gone by, I have become more and more involved in Kant’s metaphysics of nature, thanks to the work of Michael Friedman, Eric Watkins, Karl Ameriks, Dan Warren, Rachel Zuckert, Hannah Ginsborg, and Michelle Grier, among others. I think I have come to revise my views on Kant’s empirical realism too. After all, he did believe in what he called the ‘natures’ of things. His lectures on metaphysics abound with discussions about grounds and essential properties and reveal a philosopher who was no stranger to a language reminiscent of what we might call today ‘dispositional essentialism’ under the veil of the phenomenal world.

Kant barely features at all in this book, although the autobiographical journey that has led me to write it is Kantian through and through. Much as my interest in Kant’s philosophy of nature took on a life of its own between 2008 and 2014, the underlying question of realism and perspectivism continued to attract my attention. Among other philosophers of science who had directly engaged with Kantianism, four have played a key role in my journey: Margaret Morrison, Hasok Chang, Philip Kitcher, and Steven French.

I met Margie Morrison in 2007 at a conference I organized at University College London on Kant and Philosophy of Science Today. Much as she herself shared a great interest in Kant, with her unrivalled sense of humour, she used to make jokes about both Kant and perspectivism and call the latter ‘the view from everywhere’. This book owes a lot to her sharp but always light-hearted criticism. Her work on scientific models and her book Reconstructing Reality with its critique of perspectivism has provided much food for thought over the past few years. Sadly, Margie passed away just when I was finishing the edits of this book after a long illness. She is sorely missed. I cannot in all honesty claim that I have answered all her questions or appeased her criticism. But in attempting to do so, I managed to get this book project started. I had to start from somewhere, and Margie’s criticism-with-a-smile (which I discuss in Chapter 3) seemed the right place to do so.

With Hasok Chang, I share two decades of conversations that started when I was a student at LSE and continued when we were colleagues in the Department of Science and Technology Studies at UCL. With his pragmatist-tempered interest in Kantianism, C. I. Lewis, the history of science, and scientific practice, Hasok has been a fellow traveller in our HPS field from the time of the UCL conference Kant and Philosophy of Science Today to his more recent work on Realism for Realistic People (2022). My book shares his ‘integrated HPS’ approach to the topic of realism and perspectivism by indulging (probably more often than philosophers of science typically do) in historical examples and details of scientific practices.

Philip Kitcher’s work on Kant and natural kinds, real realism, and science in democratic societies has had a huge influence on me in more ways that I can possibly recount. I am grateful to him for the enthusiastic support for this book project and for the many conversations over realism, truth, progress, and pragmatism before and during the COVID-19 lockdown. Like Peter Lipton with respect to Giere’s book, he too I suspect might not be persuaded by the slight Kantian undertone of my perspectival realism. But his defence of realism in science and of the Galilean strategy behind the reliabilityof scientific knowledge is not only one of the main starting points for my journey, but also a defining feature of the kind of realism I see as compatible with perspectivism.

I cannot stress how much I have learned from Steven French’s historically informed approach to the philosophy of quantum mechanics, and the neo-Kantian Cassirer-inspired outlook of his ontic structural realism. I learned from him that answers to the epistemological question need not be outsourced to metaphysics. Or, better, such answers require a different way of thinking about realist metaphysics. He has adopted a structuralist approach. I have followed a perspectivalist one—from data to phenomena, and from phenomena to kinds. But fellow travellers we have remained through thick and thin, walking in the garden of realism in science.

Yet it is not just to my Kantian, neo-Kantian (and quasi-Kantian) fellow travellers that I owe a huge debt of gratitude. In a way, if Peter Lipton’s 2007 book review was the starting point, the launch event of this project was the PSA 2014 symposium on Perspectivism in Chicago, with Mazviita Chirimuuta, Ron Giere, Sandra Mitchell, and Paul Teller as co-symposiasts. This was the first and sadly also the last opportunity I had to meet Ron Giere in person. He was enthusiastic about the idea of a symposium on perspectivism and we celebrated afterwards in one of his favourite ‘trattorias’ in Chicago. The work on colour perception by Mazviita Chirimuuta and on semantic implications of perspectivism by Paul Teller has been inspirational and has opened up a host of philosophical questions about the role of secondary qualities and how exactly to rethink the realist question.

Sandra Mitchell’s work on integrative pluralism and laws of nature in biology has accompanied me since 2008, when I first met her at a conference in Heidelberg. She too has become over the years a wonderful fellow traveller in the land of perspectivism. I am hugely grateful to her for the very many conversations on topics as wide-ranging as the nature of phenomena, scientific models, laws, and pluralism, but also protein folding, bees, and flocks of starlings. I have learned from her to better appreciate how perspectivism does not just matter for answering the epistemological question about scientific knowledge. It equally matters for better understanding everyday scientific methodology, be it in biology or in physics or in other scientific fields.

Innumerable friends and colleagues have helped me along the way. In December 2019, I organized an away-day in Edinburgh to discuss some preliminary material from this book with Nancy Cartwright, Hasok Chang, Ana-Maria Crețu, Omar El-Mawas, Franklin Jacoby, Alistair Isaac, Casey McCoy, Mark Sprevak, and Jo Wolff. Many thanks to all of them for all the feedback and pointers that helped me write Chapter 1 and rewrite Chapter 6.

During the long months of COVID-19 lockdown in 2020, I organized a weekly reading group on Zoom to try out preliminary chapter drafts with a small group of colleagues including Julia Bursten, Ana-Maria Crețu, Adrian Currie, Joe Dewhurst, Franklin Jacoby, Catherine Kendig, and Sabina Leonelli. They were an extraordinary group in offering countless suggestions for improvements and identifying problematic passages. I cannot even begin to enumerate the number of changes I made to the manuscript after each reading group. I think the final outcome is a much better-quality book than it would have been if I had not had detailed comments and written feedback by this incredible group of friends and colleagues.

They say it takes a village to write a book. In my case, it felt more like it takes an entire city and all its surroundings, given the number of colleagues across various fields and areas whose feedback has proved important for this volume. Over the years, graduate students at the University of Edinburgh (and visiting postdoctoral fellows too) have been important interlocutors on a number of related topics: I thank Anna Ortín Nadal, Nick Rebol, Giada Fratantonio, Andrea Polonioli, Laura Jimenez, Sander Klaasse, Jan Potters, Nora Boyd, Siska de Baerdemaeker, Laura Bujalance, and Sophie Ritson. I held a number of additional reading groups in the middle of the 2020 pandemic and (in alphabetical order and I hope I am not forgetting anyone) I am hugely grateful to Marialuisa Aliotta, Helen Beebee, Franz Berto, Luigi Del Debbio, Catherine Elgin, Steven French, Roman Frigg, Peter Hawke, Sandra Mitchell, Alex Murphy, John Peacock, Stathis Psillos, Tom Schoonen, and Paul Teller for reading various extracts of the book and offering eye-opening comments.

Special thanks to Franz Berto and Timothy Williamson for helpful pointers and discussions on the material in Chapter 5 on physical conceivability and epistemic conditionals. I thank Helen Beebee, Andrew Clausen, and Wilson Poon for helpful conversations on essentialism, monetary policies, and physical transitions of state, all of which informed my discussion in Chapter 9. Craig J. Kennedy provided helpful references with kelp-making and the Scottish glass industry of the nineteenth century in Chapter 10; and Nathan Brown with chemoinformatics in Chapter 7. Conversations with Rob Rupert offered very helpful pointers for the psychology literature surrounding children’s acquisition of natural kind terms that features in Chapter 8. On ethnobiology and local knowledge, I benefited from discussions with and comments by Catherine Kendig and Alison Wylie, respectively: I learned a lot from their work on these topics.

Ken Rice very kindly gave me written comments on two different versions of the material in Chapter 4.b. on climate models, for which I also greatly benefited from comments by Roman Frigg, Benedikt Knüsel, and Wendy Parker. For Chapter 4.a, I had the privilege of receiving some really helpful references on the history of nuclear physics by Isobel Falconer and Roger Steuwer. (The latter very kindly also posted a book of his for me that I could not find anywhere in the library.) Raymond Mackintosh went over and above any possible expectation by kindly offering detailed written comments on several rewrites of the material in Chapter 4.a. I cannot stress how much I have learned about nuclear models over the past months thanks to my email exchanges with him. I also had the honour of receiving feedback from Uta Frith and John Morton, whose modelling work on dyslexia I discuss in Chapter 4.c. I am immensely grateful to all these colleagues for their generous time and help. Needless to say, and as always, any error or mistake in those three case studies in Chapters 4.a, 4.b, and 4.c and anywhere else in the book remains entirely my own responsibility.

A number of colleagues and friends across different fields over the years engaged with me in conference discussions and offered comments and pointers on innumerable related topics. I thank in particular Theodore Arabatzis, Alan Barr, Homi Bhabha, Alexander Bird, Alisa Bokulich, Abbe Brown, Matthew Brown, Jon Butterworth, Elena Castellani, Tiziano Camporesi, Panagiotis Charitos, Mazviita Chirimuuta, Heather Douglas, Marcelo Gleiser, Rebekah Higgitt, Kareem Khalifa, Michael Krämer, Marcel Jaspars, Ofer Lahav, Walter E. Lawrence, Victoria Martin, Aidan McGlynn, Tom McLeish, Angela Potochnik, Duncan Pritchard, Carlo Rovelli, Juha Saatsi, Simon Schaffer, Justin E. H. Smith, Andrew Schroeder, Mauricio Suárez, Karim Thebault, Nick Treanor, Peter Vickers, Francesca Vidotto and Jo Wolff.

But this book would not have been quite the same without the equally crucial editorial help I received all along from Jon Turney. He patiently read it all and with his eagle eyes trimmed it down and made it more readable. He helped me (and the readers) see the woods for the trees wherever the discussion was getting too cluttered with details and I myself was getting lost down one of the too many rabbit holes. Justin Dyer polished the edited chapters and offered countless suggestions for improvements in the English style as well as correcting my (incorrigible) stylistic infelicities. Peter Ohlin at Oxford University Press has been an enthusiastic supporter of this book project from beginning to end—very many thanks to him too. The staff at the National Gallery (London), Museo del Prado (Madrid), and the Brooklyn Museum (NYC) were very kind in assisting with the copyright of the images reproduced in these pages.

Some of the material here draws on articles I have published elsewhere. In Chapter 3, Section 3.2 and 3.6 are from Massimi (2018b) ‘Perspectival modelling’, Philosophy of Science 85, 335–359. Reproduced with permission from University of Chicago Press, copyright (2018) by the Philosophy of Science Association. And Section 3.3. draws on Massimi (2018a) ‘Perspectivism’, in J. Saatsi (ed.) The Routledge Handbook to Scientific Realism, London: Routledge. Reproduced with permission from Routledge. Chapter 5, Sections 5.3, 5.4, and 5.5, builds on Massimi (2019a) ‘Two kinds of exploratory models’, Philosophy of Science 86, 869–881. Reproduced with permission from University of Chicago Press, copyright (2018) by the Philosophy of Science Association. Section 5.6 expands on Massimi (2018f) ‘A perspectivalist better best system account of lawhood’, in L. Patton and W. Ott (eds) Laws of Nature, Oxford: Oxford University Press. And Section 5.7 draws on Massimi (2018e) ‘Four kinds of perspectival truth’, Philosophy and Phenomenological Research 96, 342–359. In Chapter 6, Sections 6.2 and 6.6. are reproduced from Massimi (2011a) ‘From data to phenomena: A Kantian stance’, Synthese 182, pp. 101–116. Copyright Springer Science + Business Media B.V. 2009. In Chapter 10, Sections 10.2, 10.3, and 10.4 are reproduced (in expanded form for Section 10.2) from Massimi (2019c) ‘Realism, perspectivism and disagreement in science’, Synthese (Open Access), https://doi.org/10.1007/s11229-019-02500-6. The quotes at the start of Chapters 2, 3, 5, 9, 10, and 11 are from Invisible Cities by Italo Calvino, published by Secker. Copyright © Giulio Einaudi editore, s.p.a. 1972. English translation copyright © Harcourt Brace Jovanovich, Inc. 1974. Reprinted by permission of The Random House Group Limited. For the US and Canada territories, Invisible Cities by Italo Calvino, translated by William Weaver. Copyright © 1972 by Giulio Einaudi editore, s.p.a. Torino, English translation © 1983, 1984 by HarperCollins Publishers LLC. Reprinted by permission of Mariner Books, an imprint of HarperCollins Publishers LLC. All rights reserved.

This book could only have been conceived and executed thanks to a generous and protracted period of research leave made possible by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement European Consolidator Grant H2020-ERC-2014-CoG 647272 ‘Perspectival Realism: Science, Knowledge, and Truth from a Human Vantage Point’). The same grant pays for the Open Access fees that allow you to read this book entirely for free. I will always be hugely grateful to the ERC for this unique research opportunity over the past five years, and to my ERC team—my PhD students Franklin Jacoby and Lorenzo Spagnesi, my postdocs Casey McCoy and Ana-Maria Crețu, and my project coordinators Deborah Stitt and James Collin—for the intellectual conversations, conferences, and events that made the project behind the book possible at all.

Last but not least, I owe the greatest debt of all to my family. The scattered references to Calvino’s Invisible Cities at the beginning of various chapters are in loving memory of my father, who adored Calvino and introduced me to him when I was a teenager. But they are also dedicated to my mother, who has been all along my unfailing supporter despite remaining in the village in the Sabine hills where I grew up and in COVID lockdown—by herself—for now almost two years. Writing this book has taken away far too much of my time, time that I could have spent visiting her more often.

My interest in situated and local knowledge is in part autobiographical. I grew up in a rural part of Italy. Fifty years ago, for the generation of my parents, access to higher education was still a privilege for only a few people in that part of the world. And for the generation of my grandparents in the interwar period, education opportunities stopped at the end of primary school in the best case scenario. And yet I have never encountered more reliable knowledge production about a range of natural phenomena than growing up in that local community—including my grandfather Eligio’s unfailing knowledge of olive trees flowering and pollination vital for the local rural economy and production of olive oil. It is from them—and from all local epistemic communities past and present (wherever they are)—that I derived inspiration for the discussions on local knowledge in Chapters 8–10 and the non-elitist scientific cosmopolitanism in Chapter 11.

Over the past sixteen years, my husband Mark Sprevak—being a philosopher himself—has been the most dedicated and attentive intellectual interlocutor I could have possibly had. I dedicate this book to him and to our son, Edward, who brought love, joy, and mental sanity amid the madness of finishing a monograph in the middle of a pandemic. I have treasured our countless family walks along the forking paths of nearby Craiglockhart Hill, bordered by foxgloves and wild garlic, and the uncertainty about which path to take, the steadfast climbing of the hill—on rainy muddy days and sunny ones too.