In the analysis of proxy-based control systems (or control systems in general), one needs to determine a goal for the system. However, ascribing goals to material systems introduces a set of thorny philosophical problems about how to analyze such systems objectively.
To handle this problem, the authors of the target article use teleonomy as a type of causal explanation, which may be unfamiliar to some. In systems in which purpose is reified into a mechanism, teleonomy allows us to talk legitimately about the purposes in such systems without having to decide whether a purpose is “really” there. This allows for discussions of teleology in places where purposes seem objectively clear but it is not clear that the purposes were the result of conscious choice. This was originally developed by Pittendrigh and later by Mayr to discuss the behavior of organisms, which, in their view, exhibited clear purpose-driven behavior despite those behaviors not having been designed for a purpose (Mayr, Reference Mayr1961; Pittendrigh, Reference Pittendrigh, Roe and Simpson1958). Thus, whether from accident, intentionality, or some other cause, teleonomy provides a framework to analyze purpose in reified mechanisms. Thus, the usage of teleonomy allows John et al. to discuss the structural role of purpose across a wide variety of mechanisms that do not share similarities in their origin.
However, the discussion of teleonomy, although appreciated, falls short in two main areas. The first issue is that the authors do not address how one identifies purposes objectively in teleonomic systems. Without clear methodological principles, an investigator is in danger of assigning ad hoc purposes to systems. Thankfully, the philosophical literature provides some help. Mossio and Bich (Reference Mossio and Bich2017) provide a solid means for practitioners to objectively identify at least some goals. They say that we can objectively classify a process as goal-directed if that goal requires the system to expend energy to accomplish and entails some aspect of maintaining the causal closure of the system.
The second area of concern is that, although the authors actively work with the concepts of teleonomy, they seem to be unaware of the progress made in the relationship between teleonomy and evolution over the past decade (Corning, Reference Corning2014). Although teleonomy was originally thought to not apply to evolution itself, modern work in teleonomy has actually emphasized its role in evolution (Corning et al., Reference Corning, Kauffman, Noble, Shapiro, Vane-Wright and Pross2023; Bartlett, Reference Bartlett2023). However, the perspective that the authors of the target article take on evolution is decidedly selectionist, and, although it pays some lip service to nonselectionist thinking, it does not actually incorporate any of it into the main body of work. The extended evolutionary synthesis, which has been gaining prominence in biological thought in recent years, is largely a phenomena unified by evolutionary teleonomy (Bartlett, Reference Bartlett2017). I think an analysis of proxy failure in evolutionary dynamics informed by modern approaches to evolution would be informative.
Additionally, the selectionist bent of the authors leads them to attribute too much causal power to natural selection whether in markets or ecology. Selection is an eliminative force, both in economics (market selection) and in biology (natural selection). Selection cannot create or fix any proxies either in the market or in biology. At most, it can eliminate bad proxies or reward good ones that already exist. Proxy creation requires positive action, and thus market selection cannot add to the increasing complexity of market forces.
Interestingly, the authors problematically equate opposing meanings of “selection” in Appendix 1. Natural selection is specifically at odds with intentional selection, yet John et al. equate “selection of action” with “selection of traits.” Selection of action is an intentional process where the purposeful entity selects steps toward a goal. Natural selection is an unintentional process where bad outcomes block further elaboration of failed systems. These are fundamentally different because, in selection of action, something can be selected prior to the favorable outcome, in vision of it occurring, because the functional goal is “in mind.” In natural selection, anything that is not presently functional is selected against, even if it is “on the way” to being functional, because natural selection cannot see future function, only present function.
This confusion creates ambiguity about what sort of causal mechanisms the authors are even referring to when discussing selection. In all, I think that the selectionist interpretation of evolutionary dynamics and its confusion with intentional selection has caused the authors to attribute to natural selection (both in the market and in nature) much more than it is capable of delivering.
This also opens up a question for future research. If selection is not creating proxies, what is? What are the teleonomic sources and/or requirements for proxy creation? How are levels of goals identified and created in such systems? Are lower-level goals inferred from higher-level ones (as is usually the case in conscious systems) or vice versa? How can such systems detect and accommodate proxy failures? Recognizing the function of teleonomy in creating these proxies will assist in asking the right questions, and perhaps also assist in building systems that are as robust as biological ones.
In the analysis of proxy-based control systems (or control systems in general), one needs to determine a goal for the system. However, ascribing goals to material systems introduces a set of thorny philosophical problems about how to analyze such systems objectively.
To handle this problem, the authors of the target article use teleonomy as a type of causal explanation, which may be unfamiliar to some. In systems in which purpose is reified into a mechanism, teleonomy allows us to talk legitimately about the purposes in such systems without having to decide whether a purpose is “really” there. This allows for discussions of teleology in places where purposes seem objectively clear but it is not clear that the purposes were the result of conscious choice. This was originally developed by Pittendrigh and later by Mayr to discuss the behavior of organisms, which, in their view, exhibited clear purpose-driven behavior despite those behaviors not having been designed for a purpose (Mayr, Reference Mayr1961; Pittendrigh, Reference Pittendrigh, Roe and Simpson1958). Thus, whether from accident, intentionality, or some other cause, teleonomy provides a framework to analyze purpose in reified mechanisms. Thus, the usage of teleonomy allows John et al. to discuss the structural role of purpose across a wide variety of mechanisms that do not share similarities in their origin.
However, the discussion of teleonomy, although appreciated, falls short in two main areas. The first issue is that the authors do not address how one identifies purposes objectively in teleonomic systems. Without clear methodological principles, an investigator is in danger of assigning ad hoc purposes to systems. Thankfully, the philosophical literature provides some help. Mossio and Bich (Reference Mossio and Bich2017) provide a solid means for practitioners to objectively identify at least some goals. They say that we can objectively classify a process as goal-directed if that goal requires the system to expend energy to accomplish and entails some aspect of maintaining the causal closure of the system.
The second area of concern is that, although the authors actively work with the concepts of teleonomy, they seem to be unaware of the progress made in the relationship between teleonomy and evolution over the past decade (Corning, Reference Corning2014). Although teleonomy was originally thought to not apply to evolution itself, modern work in teleonomy has actually emphasized its role in evolution (Corning et al., Reference Corning, Kauffman, Noble, Shapiro, Vane-Wright and Pross2023; Bartlett, Reference Bartlett2023). However, the perspective that the authors of the target article take on evolution is decidedly selectionist, and, although it pays some lip service to nonselectionist thinking, it does not actually incorporate any of it into the main body of work. The extended evolutionary synthesis, which has been gaining prominence in biological thought in recent years, is largely a phenomena unified by evolutionary teleonomy (Bartlett, Reference Bartlett2017). I think an analysis of proxy failure in evolutionary dynamics informed by modern approaches to evolution would be informative.
Additionally, the selectionist bent of the authors leads them to attribute too much causal power to natural selection whether in markets or ecology. Selection is an eliminative force, both in economics (market selection) and in biology (natural selection). Selection cannot create or fix any proxies either in the market or in biology. At most, it can eliminate bad proxies or reward good ones that already exist. Proxy creation requires positive action, and thus market selection cannot add to the increasing complexity of market forces.
Interestingly, the authors problematically equate opposing meanings of “selection” in Appendix 1. Natural selection is specifically at odds with intentional selection, yet John et al. equate “selection of action” with “selection of traits.” Selection of action is an intentional process where the purposeful entity selects steps toward a goal. Natural selection is an unintentional process where bad outcomes block further elaboration of failed systems. These are fundamentally different because, in selection of action, something can be selected prior to the favorable outcome, in vision of it occurring, because the functional goal is “in mind.” In natural selection, anything that is not presently functional is selected against, even if it is “on the way” to being functional, because natural selection cannot see future function, only present function.
This confusion creates ambiguity about what sort of causal mechanisms the authors are even referring to when discussing selection. In all, I think that the selectionist interpretation of evolutionary dynamics and its confusion with intentional selection has caused the authors to attribute to natural selection (both in the market and in nature) much more than it is capable of delivering.
This also opens up a question for future research. If selection is not creating proxies, what is? What are the teleonomic sources and/or requirements for proxy creation? How are levels of goals identified and created in such systems? Are lower-level goals inferred from higher-level ones (as is usually the case in conscious systems) or vice versa? How can such systems detect and accommodate proxy failures? Recognizing the function of teleonomy in creating these proxies will assist in asking the right questions, and perhaps also assist in building systems that are as robust as biological ones.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.