New Perspectives on Technology in Society

Experimentation Beyond the Laboratory

Ibo van de Poel, Lotte Asveld, Donna Mehos, Donna C. Mehos

• Publisher: Routledge
• Year: 2017
• Print ISBN: 9781032242026
• eBook ISBN: 9781315468235
• Edition: 1
• Topic: Politics & International Relations
• Subtopic: Politics
• Index: Politics & International Relations

The development and introduction of a new technology to society can be viewed as an experimental process, full of uncertainties, which are only gradually reduced as the technology is employed. Unexpected developments may trigger an experimental process in which society must find new ways to deal with the uncertainties posed.

This book explores how the experimental perspective determines what ethical issues new technologies raise and how it helps morally evaluate their introduction. Expert contributors highlight the uncertainties that accompany the process, identify the social and ethical challenges they give rise to, and propose strategies to manage them.

Focusing on the introduction of new technologies and experimentation as ways to perceive new developments and changing contexts, a key theme of the book is how to approach the moral issues raised by new technology and understand the role of experimentation in exploring these matters.

1 Control in scientific and practical experiments

Peter Kroes

Introduction

My aim in this chapter is to analyze and compare two types of experiments which I shall refer to as “scientific” and “practical” experiments. Scientific experiments are performed from a theoretical perspective; they have a purely epistemic aim and play a role in producing a specific kind of knowledge, namely knowledge about regularities in our world. Practical experiments are performed from a practical perspective; they serve practical interests and are concerned with how to act in the world in order to bring about a desired state of affairs. Whereas scientific experiments play a key role in understanding the world we live in, practical experiments play a key role in changing our world. In analyzing and comparing these two types of experiments, I will focus on the notion of control. This notion is crucial for the modern conception of scientific experiments. I will argue that the notion of control also plays an important but different role in practical experiments, and that the kind of knowledge produced in practical experiments is different from the kind of knowledge produced in scientific experiments.

I will employ a rather broad notion of the term ‘experiment’: experiments are not only performed with the aim of gathering a particular kind of knowledge about the world but also with the aim of changing the world. The first aim of experiments is generally acknowledged and traditionally associated with scientific experiments. However, many experiments performed in a technological context fall under my category of scientific experiments. Any technological experiment that is performed in order to learn regularities about how to act in or change the world is in my terminology a scientific experiment. They lead to what Hansson (2016) calls “action knowledge,” that is, knowledge of actions that if performed (adequately) in particular situations will lead to a desired result. Their aim is still knowledge of regularities, but of regularities of a special kind, namely concerning the effects of human actions. Practical experiments, by contrast, aim at bringing about a certain state of affairs in the world. The production of knowledge, be it of regularities or something else, comes subsidiary to realizing a practical result, that is, a desired state of affairs. Of course, these two aims do not exclude each other and may be combined to give rise to a whole spectrum of experiments, from experiments with a purely epistemic aim to a practical aim and hybrid forms in between.

The reason why, in the context of this book, I opt for a broad interpretation of the notion of experiment is that when the introduction of new technologies in society is conceived of as a kind of experiment, we are dealing first and foremost with a situation in which we aim at changing the world to bring about a desired state of affairs. By calling such a situation an experiment, we are drawing attention to the fact that, apart from its practical aim, we also want to learn something from it. But exactly what is it that we want to learn? And what is it that we can learn? I will argue that for scientific and practical experiments, different issues come into focus when we analyze in more detail what it means to learn from experiments, especially in relation to controlling experiments.

At first sight, my distinction between scientific and practical experiments appears to run closely parallel to Hansson’s (2015, 2016) distinction between epistemic and action-guiding experiments. Epistemic experiments aim at factual knowledge about the workings of the world, whereas an experiment is action-guiding according to Hansson if and only if the following two criteria are satisfied (2016, 617):

1 The outcome looked for should consist in the attainment of some desired goal of human action, and

2 the interventions studied should be potential candidates for being performed in a nonexperimental setting in order to achieve that goal.

The first condition of action-guiding experiments is about changing the world and is therefore very much in line with my notion of practical experiments. The second, however, shifts the focus from bringing about a particular state of affairs to learning how to bring about that state of affairs in order to be able to apply what is learned in nonexperimental settings. For Hansson, this implies that the aim of action-guiding experiments is to produce knowledge of regularities about how to act in order to achieve a goal; that knowledge may then be applied in nonexperimental settings. This is a rather strong requirement that ties experimental learning to learning of regularities and excludes, for instance, learning from experiments on the basis of analogies with nonexperimental settings. Apart from this focus on learning of regularities, there is yet another reason why action-guiding experiments are different from practical experiments. Hansson’s definition of action-guiding experiments refers to the distinction between experimental and nonexperimental settings. Although he does not elaborate on this distinction, it appears to refer roughly to a distinction between artificial, controlled, laboratory, etc. situations and real-life situations. Starting from this distinction, practical experiments are always real-life experiments because they aim at realizing a desired state of affairs in the real world and therefore have to be characterized as nonexperimental. Thus, with regard to practical experiments, the distinction between experimental and nonexperimental settings will have to take on a different meaning (see below).

Clearly, Hansson’s distinction between epistemic and action-guiding experiments is a distinction that falls squarely within what I have called scientific experiments: the aim of action-guiding experiments is primarily the production of knowledge of regularities (albeit of a particular kind), just as is the aim of epistemic experiments. Hansson analyses action-guiding experiments from a theoretical perspective because he is interested in the production of scientific knowledge. For him, scientific knowledge includes both factual and action-guiding knowledge, and therefore both epistemic and action-guiding experiments are important when studying the role of experiments in science. He observes, and I fully agree with him, that nevertheless action-guiding experiments have been almost completely neglected in historical and philosophical studies of experiments in science.

The distinction between scientific and practical experiments is more wide-ranging than Hansson’s distinction between epistemic and action-guiding experiments. Hansson’s definitions of epistemic and action-guiding experiments tie experiments in general to the production of a particular kind of knowledge, namely knowledge about regularities in the world. He emphasizes this aspect in his definition of experiments in general (Hansson 2016, 616):

…by an experiment I will mean a procedure in which some object of study is subjected to interventions (manipulations) that aim at obtaining a predictable outcome or at least predictable aspects of the outcome. Predictability of the outcome, usually expressed as repeatability of the experiment, is an essential component of the definition. Experiments provide us with information about regularities, and without predictability or repeatability we do not have evidence of anything regular.

For action-guiding experiments these regularities take more or less the form of conditional imperatives: “If you want to achieve Y, do X.” Niiniluoto (1993) refers to these conditionals as “technical norms;” knowledge of such norms constitutes action knowledge or means-ends knowledge. It is this rigid coupling of experiments in general to regularities, predictability, and repeatability that sits uncomfortably with the idea of the introduction of new technologies in society as experiments. These experiments in society are not about learning of regularities but rather about learning how to achieve a particular state of affairs by intervening in the world. Therefore, we need a broader conception of experiments but not one that is too broad: not every goal-directed intervention in the world qualifies as an experiment.

What is missing in habitual, nonexperimental, goal-directed interventions in the world to qualify as experiments is the element of learning. In particular, I will take a goal-directed intervention in the world to be an experiment if, apart from the intention of bringing about its goal, there is also the subsidiary intention to use this intervention to learn how to reach that goal. This may involve not only learning about the means to achieve the goal but also about how to adjust the goal in view of the available means. The intervention must go hand in hand with a form of systematic inquiry. In the pragmatist spirit of Dewey, this form of inquiry may be characterized as involving an indeterminate situation in which we are “uncertain, unsettled, disturbed” (1938, 105) and in which “existential consequences are anticipated; when environing conditions are examined with reference to their potentialities; and when responsive activities are selected and ordered with reference to actualization of some of the potentialities, rather than others, in a final existential situation” (p. 107). What is interesting about this view is that it connects inquiry and learning to responsive activities (interventions) and existential aspects. This learning by experimentation is not primarily of an intellectualistic type that results in action knowledge of (abstract) regularities; it is learning about how to adapt our interventions in the light of the goals pursued or how to adapt our goals in the light of the available means for achieving them. This kind of learning does not necessarily lead to action knowledge that can easily be expressed as the regularities considered above. This is because the justification of action knowledge in the form of regularities presupposes a form of experimental control that in many practical experiments is not available. In particular, I will argue that this is not the case for experiments involving the introduction of new technologies in society. First, however, it will be necessary to have a closer look at the issue of control in scientific experiments.

Control and scientific experiments

The aims of the types of experiments discussed in this section are the same: to provide information or evidence for the production of knowledge of regularities. This may, in the words of Hansson, be factual or action knowledge and therefore both his epistemic and action-guiding experiments fall under this heading. I will analyze the role of control in these types of experiments in three different domains, namely the physical sciences, the social sciences, and the technological or engineering sciences.

I will start with a brief look at the various kinds of experiments performed in the physical sciences.¹ Depending on the locus of the experiment, we may distinguish between thought experiments which take place in the mind, computational/simulation experiments in computers, and experiments in the real world; the last may take place in a laboratory setting or in the wild (I will mainly focus on laboratory experiments below). The epistemic role of thought and computational experiments has been and still is contested, one of the issues being whether or not they lead to new knowledge about the world (see, for instance (Mach 1976 (1897); Kuhn 1977)). This is not the case for experiments performed in the real world; the epistemic relevance of the outcomes of these experiments is not disputed. Another distinction, which cuts across this one, is between qualitative and quantitative experiments. Qualitative experiments show the existence of particular phenomena, such as the quantum-interference of electrons in the double slit experiment. Quantitative experiments focus on quantitative relations between physical quantities and generally make use of the principle of parameter variation (here we may think of Boyle’s experiments with gases to show the relation between pressure and volume that bears his name). Another distinction is based on the relation between experiment and theory. Depending on whether an experiment is intended to explore phenomena without the guidance of theory or is intended to test a theory, experiments may be divided, only schematically of course, into exploratory and hypothesis testing.

There are at least two reasons for performing laboratory experiments in the physical sciences:

1 experiments enable the study of spontaneously-occurring physical objects and phenomena under conditions that do not occur spontaneously in the world; and

2 conditions may be created for the occurrence of physical objects and phenomena that do not occur spontaneously in the world; these objects and systems can therefore be studied only under experimental conditions.

Thus, the experimenter creates the appropriate conditions for studying physical phenomena and objects, and may even create those phenomena and systems themselves, by creating the appropriate conditions for their occurrence. These conditions are human-made and therefore artificial; that, however, does not imply that the physical objects and phenomena themselves are human creations and thus artificial (Kroes 2003).

What I am particularly interested in here is the extent to which the experimenter has control over the system on which the experiment is performed. In this respect, there are significant differences between thought, computer, and laboratory experiments. In thought experiments, the physical system under study and the conditions under which it is studied are created in the imagination, and the experimenter has, in principle, total control over the system and conditions; (s)he is even in a position to study physical systems under conditions that cannot be realized in the world (for instance, by assuming the validity of imaginary physical laws). However, there are restrictions on what kind of systems can be studied fruitfully in thought experiments. These restrictions find their origin in the reasoning powers of the experimenter; it is, for instance, no use to perform thought experiments on systems that are so complex that it is not possible to draw any interesting conclusions about their behavior.

More or less, the same applies to computer experiments (simulations). Prima facie, the experimenter appears to have almost unlimited freedom to define the target system and its conditions. However, also in this case there are restrictions; the freedom of the experimenter is not unlimited due to constraints imposed by the computational device (computer). The computational power of the device limits the kinds of target systems that may be simulated and therefore also limits the control of the experimenter over the target system and its conditions. These limits on the experimenter’s control find their origin in the technological limits (related to hardware and software) of the computational device.

In laboratory experiments, technological constraints determine the control of the experimenter over what kind of system may be studied under what kind of experimental conditions. It is not possible to study objects or systems that contradict the laws of nature, as in thought and computational experiments. In a limited case where the scientist has no control whatsoever over the object of study and the conditions under which it is studied (e.g., the occurrence of a supernova), the scientist is dependent on Nature to perform the experiment so as to make it possible to study the system or phenomenon (see Morgan’s notion of Nature’s experiment (Morgan 2013). This does not mean that the scientist is condemned to the role of a totally passive observer and that no issues of control emerge. The observation of naturally-occurring phenomena may require the control of scientific instrumentation, for example, for measurements, during the observations.

Note that in these various kinds of experiments, the control of the experimenter stretches no further than control over the physical system studied and the conditions under which it is to be studied. The experimenter does not have control over the behavior of the system under those conditions; that is, the experimenter has no control over the outcome of the experiment.² If that would be the case, the performance of the experiment would lose much of its rationale: why perform an experiment if the outcome can be controlled and consequently constructed and predicted in advance?³ In that case, it would be difficult to explain how a scientist could learn anything about nature from experiments.⁴ Here, the theoretical perspective of these experiments shows itself: when it comes to gathering information or evidence about the world during an experiment, the experimenter is forced into the role of a passive observer, of someone who is a spectator watching nature perform its play.⁵ The role of the experiment is reduced...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• List of illustrations
• Notes on contributors
• Acknowledgements
• Introduction
• 1 Control in scientific and practical experiments
• 2 The diversity of experimentation in the experimenting society
• 3 Moral experimentation with new technology
• 4 The theatrical debate: Experimenting with technologies on stage
• 5 Social learning in the bioeconomy: The Ecover case
• 6 Cognitive enhancement: A social experiment with technology
• 7 Living a real-world experiment: Post-Fukushima imaginaries and spatial practices of “containing the nuclear”
• 8 “Dormant parasites”: Testing beyond the laboratory in Uganda’s malaria control program
• 9 Experimenting with ICT technologies in youth care: Jeugdzorg in the Netherlands
• 10 Adversarial risks in social experiments with new technologies
• Index