In this section, I propose an alternative account of levels that is fully compatible with the mechanistic framework and the way in which levels of mechanisms are construed by Craver, but which at the same time has a wider scope of application. Since this account will rely on properties as the crucial defining criterion, I shall first sketch a working definition of the concept of “properties”. In a second step, this definition of “property” will conceptually ground the alternative account of levels. Finally, I will implement both the definition of properties and that of levels into a formulation of mechanistic emergence.
What might be a working definition for the term “properties”? Inspired by Brian Ellis’ (2002) “new essentialism”[3] and Alexander Bird’s (2007) “dispositional essentialism”,[4] I propose the following view: what exist in the world are entities with individual dispositional profiles. An example that Ellis (2002) gives is the dispositions of particles to attract or repel each other. These essential dispositions, individuating the particles as that which they are, make it possible for us to formulate laws of nature. Many of these dispositions are the result of structural and organizational combinations of matter with different dispositions, e.g., ions have the disposition to form ionic crystals, which by means of the resulting structural characteristics in turn have other specific physico-chemical dispositions. Those essential dispositions alone, however, are not properties yet. That is because dispositions exist outside of an epistemic context. The property of being one meter high, for example, is dependent on the disposition of an object to exactly fit the measurement revealing it to be the height of one meter. Without the measurement, however, the property is not instantiated—only the disposition of its instantiation exists inherently in the structure of the object. Thus, according to my theory, properties are instantiated through the interaction of the essential dispositions of matter and an epistemic system. Of course, now you will ask what this epistemic system could possibly be. Admittedly, this aspect of the definition is in a particularly embryonic stage and requires further research. As a general characterization, epistemic systems are structured such that they feature sensors or gauges that capture specific dispositions of entities and provide characteristic values as an output. Human and non-human animals, as well as physical devices of measurement, are epistemic systems in this sense. Let us note the following as a working definition of “epistemic system”:
Epistemic system =Df (ES) Epistemic systems are organized (a.) such as they feature sensors or gauges that pick up a specific disposition exhibited by an entity and (b.) such as there is a transformation of that signal into a particular value characteristic of the system’s organization.
With this definition at hand we are able to formulate a working definition of properties:
Properties =Df (P) Properties are instantiated through epistemic processes, which are constituted by interactions between epistemic systems and complementary dispositional profiles of entities.
Let us now turn to how levels depend on properties. The idea here is that levels are established by the epistemic systems in use that instantiate the properties which belong to the respective level. Measuring ion conductance at an axon with electrodes, for example, establishes properties on a cytological level; whereas measuring reaction times of participants in a behavioural experiment establishes properties on a psychological level. The way in which different epistemic systems — e.g., a functional magnetic resonance imaging (f) scanner and a blood test — applied on the same entity — a human — establish different properties — a local decrease in de-oxygenated heamoglobin versus, for instance, cortisol levels in the bloodstream — on different levels — the level of brain activation versus the level of endocrine activation — shows that levels and properties are intimately connected. Coming back to our example from the beginning, different properties of the mental state compassion are instantiated in several ways: (a.) through a person as an epistemic system directed towards a myriad of dispositional interactions of her own body, which can then (b.) be picked up as values of standardized questionnaires probing those experiences while, finally, (c.) some properties of the underlying mechanism of compassion are instantiated by means of fMRI (cf. Klimecki et al. 2014). What becomes strikingly apparent in this example is that each of these different ways of property instantiation yields properties we would intuitively base on very different levels. Experience of compassion seems to have a very different quality and complexity to the more abstract numerical values of questionnaires or activation patterns visualized by fMRI. Thus, we can say that the specific way a certain property is instantiated already establishes a corresponding level. This way of defining levels offers a broader range of application than the levels of mechanisms account, since it is not restricted to properties of mechanisms but rather bears on properties in general. After these considerations, we are now in a position to put down the following as a working definition of levels:
Levels =Df (L) Levels are sets of properties established with respect to their instantiation through the same or a similar kind of epistemic system, which targets the same or a similar dispositional profile as compared to different epistemic systems targeting different dispositional profiles.
Considering the identity criteria of epistemic systems, an epistemic system identical to itself might be involved in the instantiation of two properties, that are on the same level qua being picked up by the said epistemic system—for example, a ruler instantiating the length of 1 meter and 20 centimetres. Two epistemic systems are similar if they pick up exactly the same kind of dispositions and exhibit a similar dimension of output value. For example, two rulers picking up the dispositions of a set of tables to instantiate the height of 1 meter and 59.1 inches (1.5 m) or two humans seeing the color blue. Note that the properties might be different in these cases but they are still on the same level, since they are instantiated by the same or similar epistemic system. A ruler and an infrared detector, for instance, are neither identical nor similar epistemic systems, since they differ in the kind of dispositions they pick up and have different dimensions of output values.
Having provided the two missing definitions for an account of emergence, let us now consider how these can be connected to mechanisms and implemented into a new account of mechanistic emergence. What could properties of mechanisms be? According to the definition of mechanisms given in the introduction, we have to identify entities and activities belonging to the mechanism, as well as starting and termination conditions. These might all be established by property-instantiating epistemic systems. We decompose, measure, and intervene with the phenomena or their realizing components to establish temporal, functional, or organizational properties. Properties of mechanisms on higher levels are instantiated differently to properties of mechanisms on lower levels. We can use one epistemic system to track particular sequences within a mechanism on a specific level in order to be able to recognize stepwise changes as they unfold in temporal order. For properties on higher levels, however, we have to change the kind of epistemic systems involved. We are dealing with different properties on a different level, and we cannot capture the same causal chain as in the single lower-level mechanism. Instead, we capture a synchronously emergent property on a higher level. A formulation of such an account of mechanistic emergence that incorporates the above definitions of “properties” and “levels” could be constructed as follows:
Mechanistic emergence =Df (ME) Mechanistic emergence is a special form of property instantiation, in which the novelty of the properties is established by a change in epistemic systems involved in their instantiation and through which they span a higher level compared to the level of the components, which realize the higher-level properties by means of their mechanistic organization and process dynamics, thereby changing the overall dispositional profile of the whole while, at the same time, being constrained with respect to their individual dispositions in virtue of the synchronous, non-causal constituency relation.
What is new in this account of emergence is that it acknowledges how even emergent properties ultimately arise out of perfectly explainable mechanistic processes. Unpredictability or unexplainability are no longer the defining characteristics of emergent properties themselves, but only characteristics of the epistemic context involved in their instantiation. However, these emergent properties are still novel in the sense that they are non-causal, non-aggregative products of mechanisms that come into existence only on a higher level, established through the kind of property instantiation that none of the components show in isolation.