Recent Experimental Philosophy on Joint Action: Do We Need a New Normativism About Collective Action?

Abstract

I. INTRODUCTION¹

There are two general views that social ontologists currently defend concerning the nature of joint intentional action (cf. Rachar 2021). According to ‘non-normativists’, joint action essentially requires a certain alignment of personal intentional states (e.g. Searle 2010; Tuomela 2013; Bratman 2014). According to ‘normativists’ (e.g. Meijers 2003; Alonso 2009; Gilbert 2013), joint action essentially involves normative relations that cannot be reduced to the intentional states of individuals. The most prominent normativist of joint action, Margaret Gilbert, is also the most prominent normativist of joint intention. According to Gilbert, a joint intention does not consist of coordinated psychological states but a single normative relation, which she calls ‘joint commitment’. A joint commitment is ascribed to a ‘plural subject’ or a ‘collective agent’ rather than a set of individuals. Gilbert's view entails that just as it takes more than one member of a group to generate a joint commitment, it takes all of them to terminate it. Consequently, no individual is entitled to simply leave the joint action without the permission of the others.

In two ground-breaking publications, Javier Gomez-Lavin and Matthew Rachar empirically investigate the relation between normativity and joint action in several survey studies. The conclusion both papers draw is essentially the same: people's intuitions tend to support the normativist approach to joint action, at least to some degree. They think that the more a group is thought of as closely cooperating, the more they become obligated to act in accordance with a certain shared goal or joint intention. However, in their most recent article, Gomez-Lavin and Rachar (2021) disagree with Gilbert's prediction that we have to seek permission from all participants if the individual wishes to terminate their own participation in the joint action. They argue that this intuition is not supported by their study. Instead, most of their participants seem to think that we only have an obligation to notify others if we want to leave or terminate the joint action. We therefore need, the authors conclude, a ‘new normativism’ of joint action that can explain the obligation to notify but does not require seeking permission.

In this paper, I argue that Gomez-Lavin and Rachar's study fails to establish that notifying others is sufficient to avoid ‘letting them down’, i.e. to avoid breaching the obligations generated by the joint commitment. Note that I do not merely defend Gilbert's theory of joint intention. Instead, I defend the ‘old’ normativist view that the authors attack and according to which we have a duty to seek permission if we wish to leave a joint action. First, I argue that the evidence presented does not give us a reason to give up old normativism. The main worry is that the study conflates different notions of obligation. An old normativist need not assume, for example, that we have a moral obligation to act if we do not want to. Secondly, I motivate the idea that notification to leave the joint action is not enough for counting as a normativist theory of joint action. Finally, I suggest some ideas about how to move the experimental literature on joint action forward.

II. NEW VERSUS ‘OLD’ NORMATIVISM ABOUT JOINT ACTION

I assume that an essential feature of any normativist view of joint action is that there are at least some normative pressures, e.g. obligations to do our part, that are not reducible to general moral principles or social norms, which may apply independently of any given joint action. The term ‘normative pressure’ is used here in a general sense without commitment to any view about the nature of this pressure, e.g. whether it is a moral obligation, rational requirement or something entirely different.

According to any normativist view of joint action, joint actions introduce new normative pressures. Assuming that joint actions really introduce such pressures, an interesting question that arises is how the individual may be released from it. Must we seek the permission of the others to terminate our participation in the joint action or is notifying them sufficient? I take a defining feature of a new normativism (as Gomez-Lavin and Rachar call it) to be that notification is enough to release us from the interpersonal pressure in joint activities. I take a defining feature of old normativism to be that notification is not enough and that we are obligated to seek permission to leave. Note that I here merely adhere to the labels ‘old normativism’ and ‘new normativism’ as they are used by the authors.

An old normativist view has been proposed by Margaret Gilbert (2013). According to Gilbert, joint intentions are grounded in what she calls ‘joint commitment’, i.e. the commitment of the will of a ‘plural subject’ (Gilbert 2013: 9) or collective agent (as opposed to individuals). Plural subjects and joint commitment are generated if two or more people express their readiness to commit to, e.g. uphold a decision or to intend to act in a certain way by emulating a single body. If these expressions are common knowledge between them, they have generated a ‘joint commitment’, i.e. a kind of commitment of a plural subject. This means that norms of rationality (e.g. means-end coherence, consistency), which apply to individual intentionality, also apply to joint intentionality, according to Gilbert.

As another consequence of a group's joint commitment, even if one of its members changes her individual intention, this member remains ‘obligated’ to act according to the joint commitment, i.e. to do their part in the joint action. Moreover, since the intention or commitment is a joint commitment, the individual is not just committed to act period. She is obligated toward the other participants of the joint commitment to act. It, therefore, lies in their power to release the individual from her obligations. Only if everyone expresses their willingness to terminate or change the joint commitment will the individual be released from her obligation to do her part.

For example, imagine James and Paula decide to take a walk together. To do so, they must express their readiness to jointly commit to intend (as a body) to take a walk together, such that this readiness to commit becomes common knowledge among them. This can occur linguistically as well as non-linguistically, e.g. by means of gestures or eye movements. However, the paradigmatic case is that James and Paul simply tell each other that they want to take a walk together. Once James and Paula publicly expressed their readiness to commit, and this is common knowledge among them, they now become ‘jointly committed’ to taking a walk together. This, according to Gilbert, entails that James is obligated toward Paula to keep walking with her. If he no longer intends to engage in the joint activity, the joint commitment obligates him to ask Paula for permission to stop the joint action and the joint commitment is only dissolved if both express their willingness to do so.

What, according to Gomez-Lavin and Rachar, is wrong with the old normativist picture? They argue that while some version of normativism was supported in their previous study (Gomez-Lavin and Rachar 2019), their new study (2021) actually speaks against Gilbert's view. For our purposes, the study's two main conditions High Collective Action and Promising are the most relevant.

High Collective Action

Two people are independently walking down Fifth Avenue. They spot each other at 65th street, and they walk together, chatting, laughing and maintaining their pace, until, as it happens, one of them peels off at 59th street.

Promising

Two people are independently walking down Fifth Avenue. They spot each other at 65th street, and promise to walk together to 55th street, until, as it happens, one of them peels off at 59th street.

Furthermore, the following measures are important:

1. Togetherness Measure: ‘To what extent were the two people acting together’ anchored at 0 (‘Not at all’) and 6 (‘Totally working together’).

2. Permission Measure: ‘Does the person who peels off have to seek permission to leave from the person who stays?’ anchored at 0 (‘Not at all’) and 6 (‘Totally’).

3. Notification Measure: ‘Should the person who peels off notify the other that they’re leaving?’ anchored at 0 (‘No obligation at all to notify’) and 6 (‘Total obligation to notify’).

Gomez-Lavin and Rachar (2021) first asked subjects to rate the degree of ‘togetherness’ they intuit in the above vignettes. Their hypothesis (following Gilbert) is that the more their subjects judged the joint action as exemplifying ‘togetherness’, the more would they intuit a kind of bindingness or sense of obligation. The authors find that the more the subjects intuited a kind of togetherness, the more they considered the participants of the joint action to be ‘obligated to notify’ but not to be ‘obligated to seek permission’ to leave—except in the promising condition.

The conclusion the authors draw from this is that people seem to think that the participants are obligated to notify but not to seek permission. They also found that in the promising condition, subjects do think that we are obligated to seek permission, which suggests that the promise adds additional normative pressure to the pressure already present in the joint action. This, according to the authors, supports neither Bratman nor Gilbert. Hence, they reject both Bratman's non-normativism and Gilbert's normativism and demand a new kind of normativism.

III. OBJECTIONS TO THE STUDY

There are some issues with the design of the study. First, it is not clear whether the empirical measures are sensitive enough to decide between Bratman and Gilbert. To defend his view, Bratman could simply cite social norms of politeness to explain why subjects have the intuition that we should notify others when leaving the joint action. This pressure need not be internal to any joint intention or joint action. The subjects were only asked to rate whether the people walking together are under some normative pressure to notify each other if they wish to leave. They were not asked to assess whether this pressure is external or internal to the joint action, e.g. whether it is grounded in a general social or moral norm. The latter would be compatible with Bratman's view. Of course, it might be difficult to test this question, but it would certainly be important to do so if we wish to discuss the debate between Gilbert and Bratman on empirical grounds.

Gilbert's view also does not come under pressure. First, the study conflates different notions of obligation. To understand the concern, note that Gilbert's notion is a peculiar one (but not uncommon, see Searle 2010) in that it is not a moral obligation. What she has in mind is a kind of non-moral normative pressure to act in accordance with the joint commitment that is inherent in the joint action. In their previous study, Gomez-Lavin and Rachar (2019: 115) merely showed that subjects are obligated to notify one another even if the overall action and the goal is an immoral one. In the presently discussed study, they only ask whether the one who ‘peels off’ has to ask for permission to leave. Thus, there is nothing that prevents subjects from interpreting this obligation as a moral one. So, a Gilbertian can argue that subjects might still believe that people who walk together are under another type of normative pressure to seek permission that the subjects were not sensitive to because they understood the question in a moral sense.

Another major problem is that there is a design flaw in the study: The measures are not homogenous enough to justify the conclusion drawn by the authors. The permission measure uses the expression ‘have to’, which can be understood as a kind of obligation or even requirement. Again, many subjects likely interpret this in an overly strong moral sense that an old normativism does not have to accept. The notification measure uses the term ‘should’, which can be understood in terms of a recommendation. I may believe that someone should notify another person—perhaps out of politeness—without thinking that this person is obligated to do so. Clearly, we should do things we do not have to do. I should go to the gym more often, but I am under no legal or moral obligation to do so. Moreover, shoulds are much more readily ascribed to another person than have tos. Committing myself to the idea that another person should do something is a weaker form of commitment than committing myself to another person having to do something. Both may be construed as ‘normative pressures’ but arguably pressures of different strengths and kinds.

A final problem is that the design of the study is not sensitive enough to detect a weaker kind of non-moral normativity. This normative pressure could be weak in the sense that at least for low-stake cases, like the ones tested, we are obligated (i.e. are under some normative pressure) to seek permission but that the other members are equally obligated to immediately grant it to us unless this person has a very convincing reason for their rejection. I take it that we are often in situations where we are required to seek permission but where it would be odd or even impermissible if the other person rejected our request without very good reasons. For example, think of a student who is required to seek permission to use the restroom during class. The teacher must immediately grant permission unless she has a very good reason for declining the request. This is a kind of have to that is sufficient for an old normativist like Gilbert but that is sufficiently weak such that we can doubt that many subjects are sensitive to it. Nonetheless, having to seek permission to go to the restroom in this sense is not the same as merely having to notify the teacher that one needs to use the restroom. In the former case, the teacher still has the power to reject the request even if the rejection is unjustified. In the latter case, there was no request in the first place.

Moreover, implicit public expressions of readiness to accept a request to change the joint commitment often may be enough to count as dissolving the joint intentional action. Gilbert is not committed to the idea that we have a duty to explicitly seek or grant permission. Subtle expressions of readiness can sometimes be enough. I doubt whether subjects can distinguish these subtleties without guidance. Consequently, we cannot rely on their intuitions in this particular set of experiments to reject old normativism.

Given a lack of convincing empirical evidence against old normativism, it is not clear why we should endorse a new version of normativism that focuses on notification. Such a move is not desirable for a more general reason. Obligations that we can dissolve simply by expressing the wish to do so are not really obligations in the standard sense, i.e. any sense that can motivate a normativist theory of joint intentional action. An obligation to notify that we wish to be released from an obligation to do our part in a joint action seems to be missing an essential aspect of any normativist theory of joint action. How can we still talk of an obligation to do our part if we are entitled to leave whenever we wish, as long as we make this intention public? This is especially the case if the obligations Gilbert has in mind are obligations owed to another person. If all it takes is notify you when I wish to be released from my obligation toward you, this seems to rob you of a right that you have against me.

I therefore fear that normativism of joint intentional action loses much of its bite if we can terminate a joint intention by notification only. The normative bond would be too weak if we could come and go as we please as Gilbert emphasises with her ‘concurrence condition’ on joint intentions (cf. Gilbert 2009). This would also bring normativism of joint action dangerously close to a non-normativist view. If I simply tell you that I will go home now, you might conclude that we no longer have the joint intention to, say, play chess together and simply give up. However, this does not mean that the joint intention has been cancelled by the first person as you may still be entitled to certain reactive attitudes that suggest wrongdoing (see Strawson 2008 for the notion of reactive attitudes). It would, for example, be appropriate to object by asking: ‘what about our game?’ In other words, you may still be entitled to remind me of our intention to finish the game. This suggests, following Gilbert (2013), that our notion of joint intention denotes something that remains stable even if we express our personal intention to abandon it.

Note also that obligations in joint action are similar to promissory obligations in many respects. Most normativists will agree that our obligations in joint actions are directed at other individuals (e.g. Roth 2004; Alonso 2009; Gilbert 2013). On the most prominent theories of directedness, our obligation to others depends either on their interests or their normative powers, say, to release the obligor from their obligation (c.f. May 2015). We are therefore obligated to another person unless they release us from the obligation, or their interests change. Again, on these views, simply notifying another person is not sufficient for getting released from a directed obligation. Similarly, Gilbert (2018) explains directed duties in terms of the right of the obligee to demand certain actions from the obligor, which, according to Gilbert, is grounded in their joint commitment. Again, on this picture, directed duties cannot be avoided merely by notifying the other that one is going to leave.

IV. HOW TO MOVE THE FIELD FORWARD

Empirically testing the different theories of joint action and joint intention is difficult, and I applaud the authors for their pioneering and ground-breaking work. I would therefore like to make a number of suggestions for how to move the experimental literature forward. I focus here merely on suggestions regarding the question of old versus new normativism. First, to proceed testing old versus new normativist views, we should be aware of different implications when using questions involving expressions like should and have to. The same question of whether a person should or has to ask for permission/notify could be given to different groups. One could then compare their answers and look for differences. If the difference is significant, we should be suspicious about conflating the two phrases.

Secondly, we should find a way to empirically distinguish different notions or kinds of obligation. To present similar vignettes that differ in terms of moral relevance, as done in Gomez-Lavin and Matthew Rachar (2019) was an important first step in this direction. Still, social norms may continue to apply even in immoral joint actions. However, social norms are too general to support a normativist theory of joint action, which is committed to the idea that it is the joint action that introduces normative pressures, e.g. via a joint intention. An obvious fix might be to generate a scenario where social politeness norms do not exist. We could think of a group that never developed such norms or explicitly rejects them. A prediction that is at least compatible with old normativism will be that in such circumstances, subjects may not intuit an obligation to notify (e.g. out of politeness) but still intuit an obligation to ask for permission at least in some cases. It might be difficult for subjects to imagine such a case. Intercultural research, comparing cultures with or without such social norms, might be a solution to this problem.

Thirdly, we could generate different scenarios that differ in terms of stakes. People likely have different intuitions about whether an agent has an obligation to do their part in high-stake scenarios (e.g. a scenario where a lot of money is at stake) compared to relatively low-stake scenarios, such as where people are simply taking a walk together. It would be interesting to test a number of very similar scenarios that differ only in terms of stake. The difference in stake could go from very low to low, medium, high to very high. Importantly, all scenarios should again be identified as immoral or perhaps morally neutral (as far as this is possible) by the subjects. This avoids the non-normativist objection that possible intuitions of normative pressures simply track moral obligations. Imagine, for example, a group of criminals that jointly commit to rob a small bank whose clients are mostly hard-working owners of small businesses without insurances. If subjects consistently intuit normative pressures to do one's part in immoral high-stake joint actions, then we have strong evidence for a kind of normativity that is introduced by the joint action itself. I predict that subjects will have stronger and clearer intuitions in such high-stake cases as opposed to the lower-stake cases the authors discuss.

A fourth suggestion that could be implemented in future work is to ask subjects whether they take the normative pressure observed in a joint action to be an obligation under reflection. We could also make the distinction between all things considered and pro tanto obligations explicit to them. We could ask subjects to reflect on their answer and tell us whether an agent is really obligated to the other person to walk with this person or whether it might merely be polite to do so. What if the other person does not care about politeness norms or rejects them? Intercultural research could again be helpful. Note that the emphasis on directed duties might help to further track the kind of directedness Gilbert takes to be important for duties in joint action. We could ask subjects whether a person who merely notifies another is ‘wronging’ their collaborators and whether this person owes them an apology.

V. CONCLUSION

Much more can be said about Gomez-Lavin and Rachar's most recent and very interesting experimental work on the normativity in joint action. However, the most critical shortcoming of their most recent study is that the design is not sensitive enough to really recommend a new normativism. I have argued that their measure is too coarse to differentiate between different notions of obligation and that the questions are not homogenous enough. Finally, I suggested that notification to leave may not be sufficient for a normativist theory of joint action. The normative link would be too weak if we could simply come and go as we please.²

Footnotes

REFERENCES