Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
According to some philosophers of technology, technology embodies moral values in virtue of its functional properties and the intentions of its designers. But this paper shows that such an account makes the values supposedly embedded in technology epistemically opaque and that it does not allow for values to change. Therefore, to overcome these shortcomings, the paper introduces the novel Affordance Account of Value Embedding as a superior alternative. Accordingly, artefacts bear affordances, that is, artefacts make certain actions likelier given the circumstances. Based on an interdisciplinary perspective that invokes recent moral anthropology, I conceptualize affordances as response-dependent properties. That is, they depend on intrinsic as well as extrinsic properties of the artefact. We have reason to value these properties. Therefore, artefacts embody values and are not value-neutral, which has practical implications for the design of new technologies.
Keywords: Ethics of technology, Value embedding, Artefacts, Moral value, Response-dependenceTechnology is value-laden. That is, development and deployment of technology can plausibly undermine or promote specific values (van den Hoven 2005). Moreover, beyond merely influencing what people value, technological artefacts might themselves embody values (Kroes 2012; Houkes and Vermaas 2010; Kroes and Verbeek 2014; Winner 1980). 1 Accordingly, an assault rifle like the AK47 would be morally good or bad, quite independently of how the rifle is used. While the claim about technology’s mediating role is uncontroversial, the second claim about technology’s capability to bear value is subject to considerable debate. Defenders of the value-neutrality thesis deny that technology embodies value (Pitt 2014; Peterson and Spahn 2011). Exemplifying the value-neutrality thesis, David Sarnoff, the founder of the National Broadcasting Company, claimed (cited in McLuhan 1994):
We are too prone to make technological instruments the scapegoats for the sins of those who wield them. The products of modern science are not in themselves good or bad; it is the way they are used that determines their value.
Defenders of the value-neutrality thesis concede of course that actions or events that involve artefacts can have value, but they deny that the artefact itself can have value. However, in the light of efforts to innovate responsibly, and the intuition that some artefacts, such as the AK47, are inherently more problematic than others, such as a sponge, the value-neutrality thesis is frequently denied. Some claim that artefacts have something like interactive moral agency (Verbeek 2006, 2014; cf. Peterson and Spahn 2011 for critical discussion), and therefore are not value-neutral, while others argue that artefacts can embody value independently of whether they are (relevantly similar to) moral agents or not (Flanagan et al. 2008; van de Poel and Kroes 2014; Brey 2014; Houkes and Vermaas 2010). 2 I will be exclusively concerned with the latter claim in this paper. 3
This paper aims at contributing to the project of finding a suitable account of value embedding for technological artefacts. It does so by criticizing what is arguably the most developed account of value embedding for technological artefacts, due to Van de Poel and Kroes (2014).
Van de Poel and Kroes argue that an artefact embodies a given value provided that its designed properties contribute to that value. I call this the intentional history account of value embedding (IHAVE). 4 I will show that this account raises insurmountable epistemic and metaphysical issues. Epistemically, IHAVE makes it difficult to discern whether an artefact is morally good or bad. The problem is that design intentions (to wit, the intentions that according to IHAVE partly determine whether the artefact is good or bad), are not directly perceivable. Since design intentions underdetermine design choices (to wit, the physical properties of the object), design intentions are also not indirectly perceivable. Therefore, if IHAVE were true, we would be unable to know what value an artefact embodies. Metaphysically, the problem is that embedded values cannot change if IHAVE were true. According to IHAVE, an artefact’s value is partly determined upon its creation. However, artefacts get appropriated for uses unforseen by the designer, and so their value changes perceptively. This cannot be accounted for by IHAVE. Though this is no conclusive proof that IHAVE fails, I conclude that it is a strong reason against IHAVE.
However, rather than endorsing the value-neutrality thesis, I turn to the concept of an affordance, which has its roots in ecological psychology and is now increasingly used in moral anthropology, to develop an alternative affordance account of value embedding. I argue that artefacts have affordances as relational properties. Based on a discussion of the roots of the concept ‘affordance’ in ecological psychology, and its use in anthropology, I argue that affordances are best conceived of as response-dependent properties. Finally, I argue that response-dependent properties can be finally valuable. Hence, since artefacts bear affordances, they can be finally valuable, too. This shows that the value-neutrality thesis is false, and it offers a first sketch of an account that makes sense of why this is the case.
Apart from its original contribution to the philosophy of technology, the paper’s conclusion should also be of relevance to scholars in digital ethics, many of whom are busy discussing the value of new technologies such as artificial intelligence. The affordance account of value embedding provides them with a way to make sense of the potentials and problems with new technology beyond their use.
According to van de Poel and Kroes (2014), a technological artefact embodies the value it was designed to contribute to. They give the example of a sea dyke. Suppose the dyke was designed for safety (to protect people from floods), and it actually does contribute to the value of safety (by protecting people from floods). For these reasons, they write, the dyke embodies the value of safety. More formally, they define value embedding as follows (van de Poel and Kroes 2014):
Value Embedding According to IHAVE: Artefact x embodies value V if the designed properties of x have the potential to achieve or contribute to V due to the fact that x has been designed for V.
In other words, the conditional states a necessary condition for an artefact to embody a value: the artefact must potentially contribute (in some sense) to the value and it must be doing so in virtue of being so designed. 5
The notion of value at play here as well as the notion of contributing to a value and the concept of design at play in the conditional need clarification. The kind of value that van de Poel and Kroes see embedded in artefacts is extrinsic final value (van de Poel and Kroes 2014). This is based on a distinction of value along two dimensions: the grounding of the value (intrinsic, extrinsic) and the relationality of the value (final, instrumental). An object that has extrinsic final value possesses final value (i.e. it is valuable in itself), but the fact that it is valuable in itself depends on relational properties (which make it extrinsic rather than intrinsic value). 6 For example, a wedding ring may possess extrinsic final value: the ring is finally valuable to someone, but that is only because of its relational properties (it is your ring, for example). As van de Poel and Kroes note, the existence of extrinsic final value is controversial. However, they follow Korsgaard (1983) in supposing something can possess final value even if it is not good in itself; if, for example, circumstantial factors determine whether something is good. Korsgaard gives the example of human happiness understood in a Kantian way: happiness is good for its own sake, even though happiness is not good in itself but only if brought about by goodwill (van de Poel and Kroes 2014).
The restriction to extrinsic final value is crucial for the intentional history account. It allows its proponents to say that artefacts bear value only if the properties that allow the artefact to contribute to the value are designed properties. The intrinsic features of the artefact (e.g. its physical structure) thus do not exhaustively determine the artefact’s value, but its extrinsic features, the intentions of the artefact’s designers, necessarily contribute to its value, too.
The previous point relates to the question of why the notion of design intentions is so vital for IHAVE. According to IHAVE, the artefact’s contribution to a given value does not suffice for the value to be embodied by the artefact. Instead, the artefact’s contribution to a given value must have been intended by the designer(s) of the artefact, who designed the artefact with said function in mind. They illustrate this idea by contrasting the dyke example with the example of a knife (van de Poel and Kroes 2014):
Dykes are thus designed for safety. The function of a knife is cutting; cutting of, for example, bread may be instrumental to a final value like health or survival or human-well-being. However, the attainment of such final values neither is part of the function of knives nor have normal knifes been designed to achieve such final values.
The contrast of a knife with the example of a dyke is supposed to bring out the importance of design intentions for an artefact’s embedded value: (a) dykes contribute to a value (safety) and (b) they are designed for that; hence, they embody the value safety. In contrast, though normal knives contribute to a value (health), thus satisfying (a), but they are not designed for health (thus they do not satisfy (b)), and so they do not embody that value. The importance of design intentions in IHAVE will play a central role in my criticism of the account in the sections to come.
IHAVE thus presents us with the following picture. Values are grounded in the physical characteristics of the object, which determine, as in the case of the dyke, whether the artefact contributes to the value, as well as extrinsic properties of the artefact, namely what might be called its intentional history. An artefact’s intentional history is comprised of the things the designer intended the designed thing to do (e.g. to protect people from floods, in the case of a dyke). Hence, the artefact embodies a given value in virtue of being designed to contribute to that value—the artefact’s intentional history is a necessary component in determining the artefact’s embedded value.
This sketch of IHAVE leaves open the sense in which an artefact is thought to contribute to a value. Van de Poel and Kroes do not describe the contribution relation in a formal sense, though their examples provide a reasonably good intuitive picture (think of the dyke, again). In any case, it does not matter for this paper how the contribution relation is understood.
IHAVE makes an artefact’s intentional history necessary for the question of which value it embeds. This fact gives rise to epistemic uncertainty about the values embedded in given artefacts, and it implies that embedded values cannot change. Thus, there are both epistemic and metaphysical worries with the intentional history account of value embedding.
Consider the metaphysical problems first. To see the objection consider that, according to IHAVE, artefacts will still have value in worlds without human beings because an artefact’s value does not depend in any way on its (actual) use, but on its intended use, and intended use is fixed during the design phase of the artefact. So, while humans or other intentional creatures are required to imbue artefacts with value, they are not required to uphold the embedded value. In effect, IHAVE decouples actual use of an artefact entirely from the question of whether or not the artefact embodies a value. That seems to be going too far. While an artefact’s value cannot solely be a matter of use (as the value-neutrality thesis implies), it would seem too strong to claim that the artefact’s value is not at all a matter of use. Because an artefact’s value is, implausibly, determined at its conception, I call this the Original Sin Objection against IHAVE. 7
Proponents of IHAVE might embrace the original sin objection by claiming that there is nothing wrong with an artefact’s embedded value being independent of actual use. They might point to works of art in comparison. It seems plausible, for example, to suggest that the Mona Lisa is still beautiful even if there is no one to admire the painting. There seems to be nothing in principle to block this reply. However, this reply blurs the line between value embedding in technological artefacts and other artefacts, like in artworks, and natural objects. The reply raises the question of why an account of value embedding is required for artefacts in distinction to an account of value embedding for objects more generally, including artworks and natural objects. There seems to be, however, good reason to construe value embedding differently in technological artefacts and in other types of objects. This will be so especially for proponents of the value-neutrality thesis. They insist that an artefact’s value somehow depends on how the artefact is used. If opponents of the value-neutrality thesis do not wish to deny this intuition outright, they have reason to take the original sin objection seriously.
Moreover, the original sin objection raises an issue deeper than dialectical worries. As the name suggests, an artefact gets its value at its inception or during its original design. There are two substantial problems here. First, it is a bit unclear when the value embedding takes place. Which intuitions do count and which therefore do imbue the artefact with value? If intentions contribute necessarily to the artefact’s value, then there needs to be more clarity about which intuitions count. Technically, the account assumes a three-part embodiment relation E, which can be expressed as E(F,I,V), such that a functional property F is intended (I) to contribute to value V. If in a design process there are multiple such relata, we need to know which one ends up in the object. With multiple designers, there might even be conflicts. Thus, at the very least, one would require a better understanding of whose intentions are relevant for the embedding of a value.
Even if the previous problem could be solved, there is a deeper problem still. IHAVE cannot account for relevant types of value change. Given what we called the original sin (to wit, the designers’ intention to have the artefact contribute to a value), the value thereby embedded in the artefact cannot change (short of changing the artefacts intentional history, which is impossible). So, there is a prima facie problem for the intentional history account to explain how value change is possible. To make the objection more precise, more needs to be said about the details of IHAVE as well as the type of value change in question. One type of relevant value change is the artefact changing its value; short of an argument to the contrary, it should be possible for ‘good’ artefacts’ to become ‘bad’ artefacts and vice versa. However, IHAVE allows this type of value change only by annihilating an artefact’s functional properties. Once an artefact cannot contribute to a value V anymore, it ceases to embody V. So, the only type of value change allowed for by IHAVE is the neutralization of value. There are other types of value change, however, that are relevant but which IHAVE cannot explain.
The most relevant form of value change is a change in the artefact’s embedded value. The appropriation of technology for new purposes suggests that proper value change, say, from an artefact contributing to safety to contributing to happiness, is possible. 8 Scholars studying the appropriation of technological artefacts for novel uses distinguish between three forms of appropriation: reinterpretation (a change in semantics, with little or no physical alteration, e.g. graffiti), adaptation, defined by a change in semantic and physical use (e.g. portable cassette players used to play religious sermons) that requires a violation of intended use, and finally reinvention, where semantics, use and physical structure are changed (e.g. low-rider cars). All three forms of technological appropriation suggest that an artefact’s embedded value is malleable, but since intentional history is not affected, the intentional history account cannot explain how embedded values change with change in artefacts, use and meaning.
Another form of value change concerns changes in values themselves. Suppose that values can change in the sense that evaluative properties E that now supervene on a set of natural properties N will at another point supervene on a disjunct set of natural properties N*. Roughly, what is good today might not be good tomorrow. 9 If this is possible, then IHAVE faces another difficulty of dealing with value change. When an artefact’s goodness is determined by the designer’s intentions then that can be translated as follows: the designer considers evaluative properties E to supervene on natural properties N and designs the artefact so that it fosters N. Suppose now that E supervenes on N* instead of N (and N* and N are disjunct). Then N is not E anymore but E*. However, the designer never intended anything about whatever value E* now supervenes on N. Given the centrality of intentions for IHAVE, the account cannot explain how embedded values change with change in values. Therefore, there are severe metaphysical concerns with the intentional history account.
In addition to the metaphysical problems just discussed, IHAVE faces epistemic problems. Since a given artefact’s designer’s intentions in part determine that artefact’s value, one needs to know these intentions to know the value embedded in the artefact. However, knowing an artefact’s intentional history is at the very least difficult, and, in some cases, impossible. Therefore, if IHAVE is correct, it will be difficult if not impossible to know whether a given artefact embodies some value (or any value at all). Given that the account aspires to guide engineering practice, these epistemic limitations are a severe cost. To show that an artefact’s intentional history is complicated if not impossible to determine, consider whether design intentions can be considered directly and indirectly.
Design intentions cannot be directly perceived. Inferences might be possible (which will be covered in the next section), but an agent’s intention itself cannot be perceived. Moreover, even if it were possible to perceive intentions directly, there will almost always be a temporal difference between the assessment of a given artefact’s value and the moment where the relevant design intentions were had. So, the direct perception of design intentions will likely be impossible. 10
Design intentions are difficult if not impossible to perceive indirectly. Some argue that intentions can be inferred from observed behaviour. There are two important classes of observations that may aid indirect uncover of design intentions. First, observable features of the artefact might allow inferences about intentions. However, this class of observations makes reliable inferences about design intentions unlikely. Trying to make inferences from observable, functional features of an artefact to its intentional history is fraught with error because design choices are overdetermined. A multitude of different intentions can plausibly motivate a given design choice (e.g. the use of wood for a given component of the artefact). A choice of wood as a material for a given component of the artefact, for example, can be motivated by an intention to save material costs, to source locally, to make the object smooth to touch, to avoid health-related hazards of other materials, to appeal to taste and so forth. Without a way to narrow down the path from possibly underlying intentions to a given design choice, an indirect inference to design choices will not succeed, and the artefact’s embedded value will remain epistemically opaque. Of course, such inferences might be possible in some remote cases. However, the proponent of IHAVE should argue that it is possible more often than not.
The second class of cases where design intentions may be indirectly observable is in cases where the designer communicated or even documented their design intentions to others, for example, through design briefs or in design requirements. 11 Design briefs, like documents or events, and documented design requirements, are used by designers to convey their intentions about the product’s functionality. Mainly because designers often aim to make their intentions clear in these contexts, it is plausible that somewhat reliable inferences about designers’ intentions can be drawn from them. Of course, access to such documents or event is not always possible, and documentation is not always preserved. Insofar as access to such documents is possible, it would seem that design intentions can sufficiently and reliably be uncovered (with a practical, rather than conceptual limiting factor). If that were the case, there might be hope to overcome some of the practical problems I mentioned and, consequently, my epistemic objection against the intentional history account of value embedding would lose some force. But the problem is not purely practical, as the following case illustrates. Suppose that a designer documents her intentions for some prototype P. A reproducer then reproduces the prototype so that P’s functional properties are kept constant. At the same time, the reproducer intends very different things for his replicas of P and also documents these intentions. As a result, there will be physically indistinguishable instances or replicas of P with very different intentional histories. Such situations abound closely enough in the real world. Whenever artefacts are reproduced by anyone that was not the inventor or original designer of the artefact, there will be a prototype with one intentional history, and replicas with different intentional histories, though the replicas’ physical features clearly do not necessarily differ (indeed, it is often the aim to keep these features as close to the original as possible). Existing differences between the intentions of P’s designer and its reproducer are, as IHAVE must hold, potentially evaluatively significant. But this commitment creates the epistemic problem of determining which intuitions count for making judgments about embedded value. 12 The epistemic uncertainty about determining whose intuitions count in assessing an artefact poses a significant obstacle for IHAVE. 13
These epistemic objections are not intended to establish conclusively that IHAVE fails. Perhaps there are ways for its proponents to argue that combinations of design choices enable more reliable inferences to design intentions, hence battling the epistemic imperceptibility objection. However, such work has not been done yet. 14 What is more, there seems no plausible reply to the metaphysical objections raised above. Therefore, even though this is no rebuttal of the intentional history account yet, the preceding discussion of the problems of the account provides strong reason to investigate an alternative.
It is time to introduce what I call the affordance account of value embedding as an alternative to IHAVE. The account finds empirical support in ecological psychology and, recently, anthropology and will be strengthened with philosophical distinctions. Having sketched the account and shown its empirical foundations, I will argue that it is superior to the intentional history account of value embedding.
In order to provide a full sketch of the account, let us briefly go back to the core question of the article. The core question about value embedding is: when (if at all) does an artefact possess value? The answer to this question bears practical implications. If artefacts embed values, then responsible engineers need to be careful not to embed the wrong values in their products. Several approaches aiming at responsible, ethical product development have been discussed in the literature (van de Poel 2016; Stilgoe et al. 2013). If artefacts embed values, then these approaches presuppose that there is a way for engineers to determine what values their product will embody reliably.
Here is a first gloss of the affordance account of value embedding: an artefact embodies value if and only if it enables valuable actions; the enabling of such action is what we may have reason to promote. I will now develop this gloss into a theory. To do so, I will rely on two crucial concepts: affordances and response-dependent properties. Affordances are relational properties of objects that make particular action likely, given some circumstance. Response-dependent properties are properties of objects that depend on the objects’ intrinsic and extrinsic attributes. I will show that an object’s affordances are response-dependent properties. This allows us to say that objects embed value because of their intrinsic properties while also taking into account how an artefact’s interactive effects, which depend on its extrinsic properties, contribute to its value (which the IHAVE account could not do). I will have to say more about the nature of affordances and response-dependent properties below. For now, note that artefacts can have instrumental value: they afford some action that may in itself be valuable, and the artefact is valuable in affording to bring it about. Artefacts also embody final value. Just like actions like caring for someone’s health and helping someone’s happiness can be extrinsic final values, affording safe behaviour or affording creativity can be extrinsic final values. In each case, we have reason to value (to wit, bring about or hinder) such affordances as extrinsic, final values. 15
More precisely, the affordance account of value embedding goes as follows:
Affordance Account of Value Embedding: Artefact x embodies value V iff x affords to a set of subjects S in conditions C an ability A and there is reason to positively respond to A (positive value), or there is reason to negatively respond to A (negative value).
The affordance relation will be interpreted as a response-dependent property below, and I will show that it is both an objective property of the artefact and a bearer of value. Therefore, it will follow that artefacts can embody values. To illustrate, an assault rifle like the AK47 affords subjects the ability to kill easily. This means that the AK47 is such that there is a significant probability that a subject succeeds in killing if the subject chooses to do so. Enabling killing, however, is a negative final value and we have moral reasons to oppose it. Therefore, the AK47 is morally bad.
We should now be able to see that the affordance account of value embedding fares very well against the desiderata derived from the discussion of IHAVE. To begin with, the account shows how artefacts can embody values that were not intended by the designer since affordances are (metaphysically) independent of intentions. Hence, one can make sense of various forms of appropriation and the account avoids the counterintuitive case of two physically identical objects differing in value. Second, the account makes it plausible that embedded value can be reliably detected. To determine what value an artefact embodies, one needs to determine what actions it affords. Though I am not claiming that this is always easy (compare: it is not always easy to determine what one’s abilities are), there are no principled reasons against succeeding as in the case of the intentional history account of value embedding. Third, the account makes it possible for embedded values to change. A full explanation of this feature will have to wait for another paper, but the general idea is that when either circumstances or abilities change, the artefact affords different things and thus embedded instrumental values may change, too. Moreover, when values themselves change, then what the artefact affords may cease to be finally valuable, and other affordances might become finally valuable instead. Fourth, the account explains embedded values in relation to human behaviour. The latter condition is intended to capture the intuition, presumably shared by proponents of IHAVE, that the embedding relation in the case of artefacts does relate to human use.
Given that the affordance account of value embedding satisfies several plausible desiderata, it appears to be a superior alternative to the intentional history account.
At this point, it is worthwhile to point out two routes for defending the affordance account of value embedding. The ambitious route would be to show that artefacts have value and then to show that all values are response-dependent properties. Given the ongoing axiological discussion about the nature of values, this is not the route I wish to take. Nonetheless, it should be mentioned that it leaves open, at least in principle, the claim that the values embedded in technical artefacts are response-dependent properties, and so, even the intentional history account of value embedding might ultimately be compatible with a basic claim defended by myself.
In any case, there is a shorter and no less attractive route to defending the affordance account of value embedding based on the following argument: