Johnson et al. expose a clear example of decision-making in the face of radical uncertainty that can rely on narrative practice. Their approach is all the more interesting as it has come into being entirely independent of the highly analogous work in the narrative tradition of Bruner and Vygotsky, Hutto and Gallagher, Nelson and others (Bruner, Reference Bruner1991; Gallagher & Hutto, Reference Gallagher and Hutto2008; Nelson, Reference Nelson2009; Ricoeur, Reference Ricoeur1984; Vygotsky, Reference Vygotsky1965), and thus provides an unbiased supporting case for these theories. In the following I will first briefly situate how this related work on narrative practice and narrative thinking is potentially pertinent to the authors' work, and then I will similarly review work on narrative and situation models in the context of the authors' proposal for narrative simulation. The hope is that Conviction Narrative Theory (CNT) can be strengthened by building on these foundations, and that reciprocally it can provide a case study for these narrative theories.
Under the heading of “Two modes of thought,” Jerome Bruner argued that in addition to rational thought based on logical reasoning, there is a perhaps more important and pervasive mode of human thought based on narrative (Bruner, Reference Bruner1986, Reference Bruner1987). The two modes are related to the distinction between how to know truth versus how to endow experience with meaning. One mode, paradigmatic or logico-scientific, is concerned with establishing empirical truth. The other mode, narrative thinking, deals in human intention and action and their consequences, and making meaning. Bruner attributes a primal importance to narrative in the human drive to attribute meaning to events, and speaks of narrative not only in representing, but actually in constituting reality (Bruner, Reference Bruner1991), and argues that “so our experience of human affairs comes to take the form of the narratives we use in telling about them.” This narrative mode of thought, and the need for making sense of human behavior, that is developed by Bruner is quite similar to Johnson et al.'s proposal that narrative provides an alternative to reasoning based on probabilities under radical uncertainty. Thus, Bruner identifies the distinction, and this second mode of thought that is independently proposed in the target article.
This primacy of narrative in human understanding is likewise highlighted by Gergen and Gergen (Reference Gergen and Gergen1988) who insist that stories (narratives) render events understandable and allow expectations for future events. Ricoeur (Reference Ricoeur1984, Reference Ricoeur and Wood1991) further characterizes this human predisposition to organize experience in a narrative format, in terms of agents, goals, actions, etc., the ability to encode experience in a narrative form, and to understand underlying causes in such narrative structures. This infrastructure for understanding human experience in the context of narrative should be of use in CNT.
Hutto and Gallagher exploit this notion of narrative thought for understanding human behavior in their Narrative Practice Hypothesis. This narrative practice theory holds that in human interaction, people regularly generate folk psychological narratives that explain why a person acted on a particular occasion, and that through exposure to these narratives children acquire the skills to understand and themselves produce such narratives (Gallagher, Reference Gallagher2013; Gallagher & Hutto, Reference Gallagher and Hutto2008; Hutto, Reference Hutto2007, Reference Hutto2009). Thus, theories of narrative practice hold that through normal exposure to narratives about human social interaction the child will come to learn how to interpret, react to, and respond to social contexts as provided by a theory of mind or folk psychology where psychological decision-making is made based on narrative patterns (Gallagher & Hutto, Reference Gallagher and Hutto2008; Hutto, Reference Hutto2007; Nelson, Reference Nelson2009). Again, this framework should be of interest as it supports and provides a foundation for the hypotheses in CNT.
In a related way, the authors develop ideas about how reasoning about the future can be achieved by “projecting a narrative forward” in section 7.1. While the intention is good, I believe that a more robust way to consider this would be based on the notion of situation model (Zwaan, Reference Zwaan2016; Zwaan & Radvansky, Reference Zwaan and Radvansky1998). A situation model is a mental representation of the state of affairs denoted by a narrative, rather than only a mental representation of the narrative itself. The situation model is thus a richer higher dimensional representation, and contains information that is not explicit in the narrative itself. In this perspective, the narrative is only a partial representation of the much richer underlying situation model that it describes. For forward prediction, it is thus the situation model itself that provides the basis for simulation (Zwaan, Reference Zwaan2016) and projection into the future that can be of service in decision-making.
We recently developed a model and situated robotic system that combines sensorimotor and perceptual inputs and narrative inputs to construct a situation model. Through its own experience and human narration, the system uses narrative practice to build up narrative constructions or patterns that are associated with representations of the causal and temporal structure of physical and mental states in the situation model (Mealier et al., Reference Mealier, Pointeau, Mirliaz, Ogawa, Finlayson and Dominey2017; Pointeau, Mirliaz, Mealier, & Dominey, Reference Pointeau, Mirliaz, Mealier and Dominey2021). In a proof of concept, we demonstrated how the system could use existing narrative constructions to understand (and thus potentially make decisions as required in CNT) about newly experienced situations, inspired by the narrative practice hypothesis of Hutto and Gallagher, and the notion of narrative structure of reality and perception of Bruner, Ricoeur, Gergen and Gergen, and Nelson. Again, such an infrastructure might be usefully employed in providing a basis for certain capabilities identified in Johnson et al.'s CNT.
The multi-disciplinary work described by Johnson et al. provides a rich testing ground for these theories of narrative practice. Indeed, it is quite remarkable that independent of this literature, the authors have arrived at the need for, and initial specification of, a related narrative practice system. In conclusion, Johnson et al. have arrived at the discovery of the value of narrative thinking as a way to overcome constraints imposed by radical uncertainty. It will be of interest to see if and how they reconcile their theory into the existing landscape of narrative thinking, narrative practice, and situation models.
You have
Access
Johnson et al. expose a clear example of decision-making in the face of radical uncertainty that can rely on narrative practice. Their approach is all the more interesting as it has come into being entirely independent of the highly analogous work in the narrative tradition of Bruner and Vygotsky, Hutto and Gallagher, Nelson and others (Bruner, Reference Bruner1991; Gallagher & Hutto, Reference Gallagher and Hutto2008; Nelson, Reference Nelson2009; Ricoeur, Reference Ricoeur1984; Vygotsky, Reference Vygotsky1965), and thus provides an unbiased supporting case for these theories. In the following I will first briefly situate how this related work on narrative practice and narrative thinking is potentially pertinent to the authors' work, and then I will similarly review work on narrative and situation models in the context of the authors' proposal for narrative simulation. The hope is that Conviction Narrative Theory (CNT) can be strengthened by building on these foundations, and that reciprocally it can provide a case study for these narrative theories.
Under the heading of “Two modes of thought,” Jerome Bruner argued that in addition to rational thought based on logical reasoning, there is a perhaps more important and pervasive mode of human thought based on narrative (Bruner, Reference Bruner1986, Reference Bruner1987). The two modes are related to the distinction between how to know truth versus how to endow experience with meaning. One mode, paradigmatic or logico-scientific, is concerned with establishing empirical truth. The other mode, narrative thinking, deals in human intention and action and their consequences, and making meaning. Bruner attributes a primal importance to narrative in the human drive to attribute meaning to events, and speaks of narrative not only in representing, but actually in constituting reality (Bruner, Reference Bruner1991), and argues that “so our experience of human affairs comes to take the form of the narratives we use in telling about them.” This narrative mode of thought, and the need for making sense of human behavior, that is developed by Bruner is quite similar to Johnson et al.'s proposal that narrative provides an alternative to reasoning based on probabilities under radical uncertainty. Thus, Bruner identifies the distinction, and this second mode of thought that is independently proposed in the target article.
This primacy of narrative in human understanding is likewise highlighted by Gergen and Gergen (Reference Gergen and Gergen1988) who insist that stories (narratives) render events understandable and allow expectations for future events. Ricoeur (Reference Ricoeur1984, Reference Ricoeur and Wood1991) further characterizes this human predisposition to organize experience in a narrative format, in terms of agents, goals, actions, etc., the ability to encode experience in a narrative form, and to understand underlying causes in such narrative structures. This infrastructure for understanding human experience in the context of narrative should be of use in CNT.
Hutto and Gallagher exploit this notion of narrative thought for understanding human behavior in their Narrative Practice Hypothesis. This narrative practice theory holds that in human interaction, people regularly generate folk psychological narratives that explain why a person acted on a particular occasion, and that through exposure to these narratives children acquire the skills to understand and themselves produce such narratives (Gallagher, Reference Gallagher2013; Gallagher & Hutto, Reference Gallagher and Hutto2008; Hutto, Reference Hutto2007, Reference Hutto2009). Thus, theories of narrative practice hold that through normal exposure to narratives about human social interaction the child will come to learn how to interpret, react to, and respond to social contexts as provided by a theory of mind or folk psychology where psychological decision-making is made based on narrative patterns (Gallagher & Hutto, Reference Gallagher and Hutto2008; Hutto, Reference Hutto2007; Nelson, Reference Nelson2009). Again, this framework should be of interest as it supports and provides a foundation for the hypotheses in CNT.
In a related way, the authors develop ideas about how reasoning about the future can be achieved by “projecting a narrative forward” in section 7.1. While the intention is good, I believe that a more robust way to consider this would be based on the notion of situation model (Zwaan, Reference Zwaan2016; Zwaan & Radvansky, Reference Zwaan and Radvansky1998). A situation model is a mental representation of the state of affairs denoted by a narrative, rather than only a mental representation of the narrative itself. The situation model is thus a richer higher dimensional representation, and contains information that is not explicit in the narrative itself. In this perspective, the narrative is only a partial representation of the much richer underlying situation model that it describes. For forward prediction, it is thus the situation model itself that provides the basis for simulation (Zwaan, Reference Zwaan2016) and projection into the future that can be of service in decision-making.
We recently developed a model and situated robotic system that combines sensorimotor and perceptual inputs and narrative inputs to construct a situation model. Through its own experience and human narration, the system uses narrative practice to build up narrative constructions or patterns that are associated with representations of the causal and temporal structure of physical and mental states in the situation model (Mealier et al., Reference Mealier, Pointeau, Mirliaz, Ogawa, Finlayson and Dominey2017; Pointeau, Mirliaz, Mealier, & Dominey, Reference Pointeau, Mirliaz, Mealier and Dominey2021). In a proof of concept, we demonstrated how the system could use existing narrative constructions to understand (and thus potentially make decisions as required in CNT) about newly experienced situations, inspired by the narrative practice hypothesis of Hutto and Gallagher, and the notion of narrative structure of reality and perception of Bruner, Ricoeur, Gergen and Gergen, and Nelson. Again, such an infrastructure might be usefully employed in providing a basis for certain capabilities identified in Johnson et al.'s CNT.
The multi-disciplinary work described by Johnson et al. provides a rich testing ground for these theories of narrative practice. Indeed, it is quite remarkable that independent of this literature, the authors have arrived at the need for, and initial specification of, a related narrative practice system. In conclusion, Johnson et al. have arrived at the discovery of the value of narrative thinking as a way to overcome constraints imposed by radical uncertainty. It will be of interest to see if and how they reconcile their theory into the existing landscape of narrative thinking, narrative practice, and situation models.
Financial support
This research was supported by the ANR France Relance; and Région Bourgogne-Franche-Comté ANER Robotself.
Competing interest
None.