Strong & Weak Emergence
Emergence describes how new higher level properties, patterns and functionality form as we put component parts and systems together. Emergence is a pervasive phenomenon in our world exhibited by virtually all types of complex entities, such as plants, animals, human societies, cultures, and economies. It is apparent that new levels of organization are formed as we put things together. When we look at a flower, we do not see a composition of molecules, cells, tissues and organs but in fact a whole flower, a corporation is given characteristics and legal rights that are not associated with any of its members. Although this emergence of patterns on different levels is apparent, what is not apparent is whether these emergent organizations are fully determined and understandable with reference to their basic component parts. Or do they exist in some why independently from the elemental that constitutes them; which would mean that they can not be entirely reducible to causal accounts derive from the elemental parts.
This distinction within emergence is captured in the terms strong and weak emergence. Where weak emergence refers to the idea that even though emergent phenomena are unexpected given the principles governing the low-level domain, they are still fully explicable only with reference to lower level phenomena.1 In contrast a high-level phenomenon is strongly emergent when it arises from the low-level domain, but facts concerning that event are not derivable even in theory from the features of the low-level domain.2
Weak emergence is the phenomenon whereby new and unexpected patterns emerge due to the interaction between the parts, the sheer number of parts and interactions makes it computationally extremely difficult to compute based upon elementary parts and their rules. However, from the weak emergence perspective, given sufficient computational capability, it would be possible to derive the higher level phenomena and thus they are seen to be – theoretically – a derivative. New phenomena and patterns do emerge, but the reason we can not predict them is that the number of interactions between components of the system increases exponentially with the number of elements and we typically do not have the computation capabilities to deal with this.
With weak emergence, it is possible to compute the high-level phenomena but typically much easier just to look at it directly. Thus this weak emergence can be understood as a kind of explanatory emergence. That is to say, the emergent features are ontologically and causally derivative but in practice, they are explanatorily irreducible due to computational complexity. New rules may appear to emerge at the different levels but ultimately if we had the computational capabilities we would be able to understand all of the rules at the different levels with respect to the lower level rules. Thus if you had an extremely high level of computation capability you would not need to focus on the higher level phenomena but understand it from first principles; one could look at it as caused by purely discrete cellular parts.
A good example of weak emergence is a cellular automaton computer program like The Game of Life.3 The Game of Life is played on a grid of checkers where a cell can be either on or off; there are four simple rules as to whether a cell should be on or off depending on the state of it immediately surrounding neighbor cells. These simple rules, when computed, can create very complex and subtle emergent patterns that appear to have their own internal structure. Such as blinkers where a group of cells “blink” on and off or gliders that seem to glide across the screen all of which are emergent phenomena. The program exhibit sensitivity to initial conditions and it is very difficult to predict what will emerge based on the initial conditions and ground rules.
Although these programs can create emergent patterns they are said to be weakly emergent because they are determined by the elementary rules, the starting state and because there is no downward causation; the macro level system does not change the micro level rules. This weak emergence is characterized by the interaction between parts as the system evolves leading to computational complexity and the appearance of something new emerging, when in fact, it is theoretically reducible to causal accounts of the elementary parts. One can not in any straightforward way derive the high-level phenomena from the fundamental rules alone. Thus compact representation – such as equations – do not tell us very much of what is going on because we need to compute the interactions to produce the high-level phenomenon.4 These weakly emergent higher level phenomena do not affect the lower levels i.e. there is only upward causation present, the macro level is determined by the micro, but not vice versa. There is an asymmetrical flow of determination, macro level patterns are not doing anything over and above what the micro level events are doing to affect the positions and behavior of the elementary parts.
An event is thought to be strongly emergent when the high-level phenomenon derives from low-level events, but a complete description of the emergent pattern is not reducible, even in principle, to an account of the elementary parts and their interactions.5 Along with irreducibility, downward causality is commonly cited as a criterion for strong emergence.6 Strong emergence entails the idea that something truly new emerges at the different levels of organization that can not theoretically be reducible to accounts of the elementary parts. The whole is something truly other than the parts. Thus it makes sense to talk about qualitatively different levels or dimensions to the system as the rules that apply on one level become replaced – at least partially – by rules of a qualitatively different nature on another level. These higher level patterns then can exert a downward cause on their constituent parts affecting their structure and functioning. Strong emergence describes the direct causal action of a high-level system upon its components; qualities produced this way are irreducible to the system’s constituent parts.7
One of the classical examples of strong emergence given is quantum entanglement. Quantum entanglement is a phenomenon within quantum physics where two particles spin states become “entangled” meaning the state of one is entirely dependent on the state of another. It has been empirically proven that the combined “entangled” organization determines the spin direction of the parts. The two particles can be lightyears away from each other but if the spin is changed on one this will be immediately reflected in a change in spin in the other. Thus the combined organization is in some way affecting a downward causation on the parts.
Another example from physics of strong emergence is water, being apparently unpredictable even given a meticulous analysis to the properties of its constituent atoms.8 It would appear that no computational description of the system can exist, for such a simulation would itself constitute a reduction of the system to its constituent parts.9 The emergent phenomenon, in this case, can not be described with reference only to fundamental rules but requires some form of macro level rule. Likewise, consciousness is another often cited example of strong emergence.
Whereas with closed systems the whole should be theoretically derived and caused by the parts, this, however, may not be the case in open systems. The parts form the whole but then the system has to interact with its environment as an entire system – not as a set of parts. This interaction requires that it perform particular functions and activities as a whole, such as walking which can only be achieved by a combination of two legs being coordinated and interdependent. For the system to operate within the environment as a whole it has to exert a downward effect on the parts in order to coordinate them towards performing macro-level processes that are required to interact and respond to the environment.
For example, we can derive the internal workings of a truck from the basic laws of engineering and physics. But using those same elementary internal rules we would never be able to derive why it is designed to drive on the left or right-hand side of the road. This phenomenon is not a product of the internal logic of the truck’s design or of the physical laws that govern its workings. It is instead a product of the system’s interaction with other systems and some historical, political and economic contingent within the environment; which may be seen to be exerting downward causation on the design of the truck. The biologist Peter Corning illustrates this when he writes: “the debate about whether or not the whole can be predicted from the properties of the parts misses the point. Wholes produce unique combined effects, but many of these effects may be co-determined by the context and the interactions between the whole and its environment.”
Types of Inquiry
This question of strong and weak emergence is of major relevance as it goes a long way to defining whether we should focus our inquiry on the micro level structure and rules that give rise to the high-level phenomena; as would be the case if we posit that the world is weakly emergent. Or whether we should instead focus on the internal patterns of the emergent phenomenon, in their own right – as would follow naturally from a belief in strong emergence.
As such the concepts of strong and weak emergence form part of the foundations within the different scientific paradigms of reductionism and holism. Reductionism is based on the premise that complex phenomena can be broken down into simple “building blocks” from which high-level events can be reconstructed. Thus a reductionist approach would typically ascribe to a weak emergent view of the world. Where complex macro level phenomena would admittedly take vast amounts of computation to derive from the basic physical building blocks but, given such a capability they would be fully explicable from the elementary parts and rules. The weak emergent theory inspires the idea that the goal of science is to understand the basic building blocks and rules for their combination and drives a quest for the “theory of everything” as seen to be found in some elementary particles; as is currently the quest of such approaches as string theory.
Systems thinking is instead interested in patterns and processes, it refers not to building blocks but more to patterns of organization and processes of change that are common to all types of systems on all scales without interest in reducing higher level phenomena to those of a lower level. As such the systems approach is built on a strong emergent view of the world. Given strong emergence, a “theory of everything” derived from elementary particles – such as strings – would end up being just one of many components necessary for a complete understanding of the universe and thus not necessarily the only one.10 Within the systems paradigm, as all phenomena can not be simply reduced to an account of elementary parts, the goal of providing a unified description of the world is instead looked for in abstraction. Systems thinking looks at how these emergent patterns on different levels have similar dynamics and from this tries to develop abstract, generic models that are relevant to all scales because they capture the features inherent to emergent processes on all level.