Emergence is a slippery and confusing concept, one that has the capacity to produce endless debate (see Taylor 2014 for a useful and enjoyable summary). In this post, I will claim that the concept of “strong emergence” as proposed by Chalmers (2006) is incoherent and, by implication, not very useful.
Yes, you read that right. I am indeed pulling Chalmers’ leg. On one hand, much of his work relies on metaphysical “possibility”, which requires coherence; on the other, his work on consciousness is facilitated by the presupposition that strong emergence exists, at least in the case of consciousness.
This second pillar of Chalmers’ approach is explicitly tackled in “Strong and Weak Emergence” (2006), which is what I’ll use as my starting point.
His definition of weak emergence is:
We can say that a high-level phenomenon is weakly emergent with respect to a low-level domain when the high-level phenomenon arises from the low-level domain, but truths concerning that phenomenon are unexpected given the principles governing the low-level domain.
Strong emergence is (initially) defined as follows:
We can say that a high-level phenomenon is strongly emergent with respect to a low-level domain when the high-level phenomenon arises from the low-level domain, but truths concerning that phenomenon are not deducible even in principle from truths in the low-level domain.
This definition is then used to propose that yes, strong emergence does indeed exist, because we know of one clear instance of it: consciousness.
even if consciousness is not deducible from physical facts, states of consciousness are still systematically correlated with physical states. In particular, it remains plausible that in the actual world, the state of a person’s brain determines his or her state of consciousness, in the sense that duplicating the brain state will cause the conscious state to be duplicated too. That is, consciousness still supervenes on the physical domain.
What Chalmers is pointing at (supervenience) is often considered a requirement for “strong emergence”: given how the universe works, an emergent property of a given system is determined by the state of the system. Change this state appropriately, and the property changes or disappears. The phenomenon then qualifies as strongly emergent if, even if we do know all there is to know about the state of the system (its structure and the lawful behaviours of its parts), it is still impossible for us to deduce the appearance and properties of the emergent phenomenon, even in principle.
I must confess that I do suspect that I’m misunderstanding Chalmers’ claims, because in my view, he quickly and directly undermines his own argument by stating:
If there are phenomena that are strongly emergent with respect to the domain of physics, then our conception of nature needs to be expanded to accommodate them.
This suggests that the lawful connection between physical processes and consciousness is not itself derivable from the laws of physics but is instead a further basic law or laws of its own.
[…] I think this account provides a good general model for strong emergence.
[…] In any case like this, fundamental physical laws need to be supplemented with further fundamental laws to ground the connection between low-level properties and high-level properties.
Let’s unpack this a little. Cases of strong emergence imply that we have a known “system” and a complete map of its internal parts, including all laws governing the interactions between them. Having this, there are still properties or phenomena of this whole system which we can’t deduce from our existing knowledge. The interesting part is Chalmers’ suggestion about what we should do next: we need to “supplement” additional laws which would specify how high-level properties supervene on low-level ones. Implicitly, it seems to me that Chalmers is claiming that, if we’ll succeed in doing so, our job will be done, and we would have “explained” our strongly emergent phenomenon.
What puzzles me is that, if indeed we will manage to achieve what Chalmers suggests, we would have concurrently demonstrated that we were dealing with a case of weak emergence. After all, our final epistemological situation is that we have “discovered” some new “fundamental laws”, which in fact allow us to deduce “emergent facts” given our knowledge of the system state, its parts and the laws that describe how these parts interact.
It seems to me that Chalmers is claiming that the correct way to handle cases of strong emergence is to admit our ignorance and start looking for fundamental laws that we haven’t yet discovered. I have no problem with this, but I also don’t see how it can be compatible with Chalmers’ own definition of strong emergence.
In other words, I think that Chalmers’ own reasoning is incoherent, and very manifestly so . Which is puzzling, because the likelihood that Chalmers may commit such macroscopical mistakes is minute. Thus, my mind immediately generates a follow up question: what could possibly explain how such a mistake could go undetected?
My answer to this question points to the interaction of two sources of error:
- A misunderstanding of how scientific progress happens.
- The incorrect assumption that “levels” (the distinction between a system and its components) are ontologically fundamental, while they are entirely epistemological.
Dealing with the first mistake feels extremely easy to me (this usually indicates I’m wrong). Scientific discovery, in my view, proceeds exactly as Chalmers proposes: we start from a point where we think we have our full picture. We claim that we understand all properties and relations that occur within a system. But in fact there is a handful of additional properties that we can’t yet account for. So we accept our ignorance, scratch our heads, think, experiment some more and eventually figure out what we were missing. Sometimes this requires the formulation of new fundamental laws (think of electro-magnetism, or of the laws regulating the relations between speed, time and mass). How we can find and formulate these additional laws is something that typically changes from case to case, but I do not think that it’s controversial to claim that for a good number of “previously unresolved” scientific problems, we eventually did.
What I’m observing here is that every time we do, we also demonstrate that what initially looked as a case of strong emergence (if applicable) was in fact weak emergence. It looked like strong emergence, because we had no knowledge of some missing piece, which in some case might be best described as a new “fundamental law”. If I’m right, this implies that Chalmers is claiming that consciousness is, hopefully, a case of this kind, which however is not what Chalmers is explicitly claiming… (Interim conclusion is that I must be missing something, but I don’t know what it is, since I’m missing it!)
The second source of confusion is about levels. I think that perhaps, with a fair amount of effort, it is possible to unpack what Chalmers claims in a coherent way, if we start from the assumption that the distinction between a system and its components isn’t arbitrary. Accepting this view would allow us to create a sharp distinction between the “level” at which “the fundamental laws of physics” apply, and a separate one that concern “fundamental psychophysical laws”. Looking at this in detail, however, uncovers a problem: the levels of explanation that are used in science are manifestly epistemic. We decide what’s a system and what are its part in an entirely opportunistic way, depending on what it is that we’re trying to achieve (our methods are always somewhat reductionist, because that’s how our own minds work). On the other hand, everyone knows that nature does not do any such distinction: some animals have sensory capacities that rely on weird quantum effects, for example. I think that it is self evident that natural phenomena operate on all possible levels “at once” and that indeed, we begin trying to understand how nature operates by “slicing it up” into different levels, picking our distinctions based on the regularities that they allow to uncover.
Put in another way, the distinction that Chalmers picks up, between fundamental laws of physics and fundamental laws of psychophysics is a distinction that refers to how we conceptualise reality and not to how reality works. This in turn may generate some confusion and lead to the idea that something can be a case of strong emergence in terms of fundamental physics, but is otherwise explainable with the aid of ontologically distinct laws of psychophysics. However, it is self evident (to me!) that if something is “strongly emergent” from one specific point of view, but is explainable from another, then the phenomenon in question cannot be considered “strongly emergent” in an all-encompassing metaphysical sense.
[Side note: assuming that it is possible to reconstruct someone else’s mental processes leading to what seems to be, in my eyes, a mistake, is an act of extreme arrogance that I am not comfortable with. I’m doing it here because one of my purposes is to try to identify what it is that I’m missing, and I cannot do so without exploring my own reasoning in full.]
Be as it may, I find myself forced to conclude that indeed, strong emergence, as described by Chalmers is a concept that fails to point to anything that is properly conceivable. Once one does, as in Chalmers’ case, include the “supervenience” side (given a state of the system, it will be necessarily associated with the emergent phenomenon), it follows that such situation implies our current inability to deduce the emergent phenomenon, given what we know of the system, and therefore immediately suggests that we do not know everything that there is to know about it.
This leads me to a final consideration, which departs from Chalmers’ view, but I think is nevertheless useful to enunciate. I’ve mentioned above the opportunistic nature of scientific theorising: levels of analysis and theoretical frameworks are ultimately picked on the basis of how useful they prove to be, not on the basis of some stable and well understood ontological principle. For this reason, when I find myself criticising a given concept I think it’s necessary to also ask the following question: irrespective to what it refers to and regardless to how concrete this referent is, is this concept useful in one way or the other?
My answer for the case of strong emergence is: yes, the concept is somewhat useful, but for one very limited reason only. The reason is that in practice, whenever something looks like a strongly emergent phenomenon, we can and probably should deploy a modicum of induction. We’d then realise that the history of science can be summarised as a series of repetitions, following a broadly repetitive pattern. We start from a situation where it feels like we have learned everything there is to know about a given subject, except for one or few secondary aspects, and some of which, upon close inspection, look like strongly emergent: given our otherwise “complete” knowledge, we still can’t explain their existence. Naturally, people will start concentrating on those few outliers and eventually someone will propose new conceptualisations, or perhaps design new instruments that allow to measure things we didn’t even know existed. These new advances will be promoted as “superior”, specifically because they allow to explain also one or more of the previously mysterious emergent phenomena.
The important side effect of realising this is practical: even if we assumed that strong emergence can exist, whenever we are presented with a situation that suggests strong emergence, the only thing we should do is to proceed as if we were dealing with a case of weak emergence, accepting that the appearance of strength is usually a function of our own ignorance.
We would thus respond by looking for alternatives ways to conceptualise and analyse the phenomenon at hand, or, if you prefer, we would redouble and renew our theoretical efforts, explicitly searching for new clues about the mechanisms we don’t fully understand. The alternative isn’t viable: should we accept something as a strongly emergent thing, we would be implicitly declare it as unexplainable, even in principle, and I fail to see any reason why we would want to.
This points to a twisted kind of negative usefulness for the concept of strong emergence: labelling something as such is functionally equivalent to declaring it “inexplicable, even in principle”; however, doing so also shows that we do not posses the intellectual tools required to reliably identify something “strongly emergent”. From our limited epistemic position, we have no way to distinguish between strong and weak emergence (assuming that strong emergence is indeed a coherent concept), and therefore we should always behave as if all emergence is weak. Failing to do is equivalent to giving up, which happens to be the exact opposite of what both science and philosophy try to achieve.
Notes and bibliography
 In case you’ll enjoy reading the present post, I would highly recommend to also read Taylor’s article. It is very well written and will provide a good description of the landscape. It will also count as a much needed correction to all the over-simplifications that I’ve scattered in here.
 It’s important to note here that I am not claiming that all possible formulations of “strong emergence” are necessarily incoherent. I’m claiming “only” that what Chalmers proposes in his paper is. Specifically, his treatment of the one (supposedly) known case of strong emergence implies it is not (or that it hopefully isn’t) a case of strong emergence.
 An important disclaimer is due: the “reductionism” concept is, just like emergence, a slippery and ambiguous one. In this case, I am pointing to a methodology and not a metaphysical stance. I claim that scientific understanding always relies on “slicing up reality” in some way, and that this happens because it is how human cognition operates. It is entirely possible to proceed in this way without assuming an all-encompassing reductionist metaphysics, and indeed, in my opinion, that’s probably what we should do.
 To be fair, it should noted that the concept of emergence does in itself rest on the assumption that different levels exist. For those of us who regard emergence as “obviously” epistemic (as I do), this detail is crucial. Given that the different levels are in our minds and not in the world out there, “emergence” becomes the side effect of how cognition works. On the other hand, if one assumes that the different levels have an ontological status, then perhaps a more metaphysical view of emergence may start making more sense (ignoring what to me looks like a foundational mistake).