Hierarchical versus network coherentism
The Stanford Encyclopedia of Philosophy is a valuable internet resource. I here refer to its article on coherentist theories of justification.
The article says:
The above characterisation of coherentism must be understood as very weak. Specifically, it is really only the negation of foundationalism:
If we understand coherentism in this weak sense, then clearly subsumptive justification is a form of coherentism. However, note that because coherentism is so broadly construed, radically different systems of justification fall under it. Some forms of justification I would approve of, and some I would not. Specifically, I want to distinguish hierarchical from network systems of justification.
A hierarchical system of justification involves a strict hierarchy of claims. Each claim has its own level in the hierarchy that derives from a lower-level claim in the hierarchy. The only thing that makes this system different from foundationalism is that the chain of justification does not necessarily follow the entailment order of the claims. Sometimes the derived claims will be justified in terms of the fundamental claims. But sometimes the fundamental claims can be justified in terms of the derived claims. In fact, the central point of justification of a whole chain of beliefs might well be somewhere in the middle of the chain. Below that central point, the justification follows subsumptive order and above that central point, the justification follows normal entailment order.
Contrast this with a network theory of justification. Here, no strict chains of belief would be found that followed any order involving entailment, either forward or reverse. Instead, each claim would contain its own set of links of justification to some other claims. Some claims would contain a greater number of such links to or from them than other claims, and some claims would contain stronger links of justification than other claims. It is analogous to the network database of HTML pages on the world wide web, where no pages are foundational. Every page is simply a node in a network, some nodes of which are more important than others purely because they receive more traffic. The equivalent of an important claim in a foundationalist theory is one where many different claims are justified in terms of it.
The clear advantage of a network database on the internet is that it makes for robust navigation via hypertext. If one of the pages is inaccessible, it can be bypassed quickly so that navigation can occur around it. The overall traffic of the network is therefore preserved. Analagously, a coherentist theory of justification can have the advantage that one's system of beliefs becomes much more robust. If one belief becomes questionable, the other beliefs can still hang together well because of their coherence with the other beliefs in the system. The entire system of belief does not collapse just because one or two claims in it are found to be faulty. Most beliefs in the network will rely on justification from more than one belief, and so they will not be jeopardised just because any one belief is thrown into question.
This all sounds very sophisticated and modern in theory, but it does not seem utilised in practice. Science and mathematics both rely on hierarchical justification. Even the assumptions of a paradigm form foundations for derived assumptions relating to the methods of hypothesis, experimentation, confirmation, refutation and so on. Allowing some of the assumptions to be justified in terms of the fruit they bear makes for some coherentism. But this is coherentism that preserves a rigorously hierarchical nature. Science has always been constructed like a building, with a foundation that allows for any edifice that might be erected upon it. Geometry provides a clear example of how sophisticated theorems are built up step by step from fundamental axioms.
In fact, it is hardly clear exactly what a network justification of science would even look like. But it certainly sounds chaotically unstructured on the face of it. The reason that science utilises a hierarchical system of justification is that it is easy to understand, debug and maintain. Coherentism allows for a greater flexibility in how chains of entailment can be justified, but to get rid of the chains would appear to be a grave mistake. Right now, if someone wants to question a particular scientific belief, the exact point of attack seems very clear. One can pinpoint exactly which link in the chain of entailment is weakest and strike there. If each claim in science had multiple points of justification, it would be much harder to arrive at conclusive refutation, which is an essential part of scientific progress, as Karl Popper recognised.
Therefore, coherentism may be admitted as a more fruitful explanation of the power of science. But it must be very clear what kind of coherentism is being implied. It is hardly an amorphous network of beliefs that are difficult to deny because of their multiple points of 'confirmation'. It would be very difficult to evolve such a system the way that science must evolve, because the very line between confirmation and refutation would become decidedly blurred. It would never be clear exactly when a hypothesis had been refuted because it might always link up to another claim somewhere else. If we keep a rigorous chain of entailment, however, science is far more capable of deciding between competing claims. Hence, a hierarchical coherentism must be supposed.
The article says:
Coherentists typically hold that justification is solely a function of some relationship between beliefs, none of which are privileged beliefs in the way maintained by foundationlists, with different varieties of coherentism individuated by the specific relationship among beliefs appealed to by that version.
The above characterisation of coherentism must be understood as very weak. Specifically, it is really only the negation of foundationalism:
Coherentism thus claims, minimally, that not all knowledge and justified belief rest ultimately on a foundation of noninferential knowledge or justified belief.
If we understand coherentism in this weak sense, then clearly subsumptive justification is a form of coherentism. However, note that because coherentism is so broadly construed, radically different systems of justification fall under it. Some forms of justification I would approve of, and some I would not. Specifically, I want to distinguish hierarchical from network systems of justification.
A hierarchical system of justification involves a strict hierarchy of claims. Each claim has its own level in the hierarchy that derives from a lower-level claim in the hierarchy. The only thing that makes this system different from foundationalism is that the chain of justification does not necessarily follow the entailment order of the claims. Sometimes the derived claims will be justified in terms of the fundamental claims. But sometimes the fundamental claims can be justified in terms of the derived claims. In fact, the central point of justification of a whole chain of beliefs might well be somewhere in the middle of the chain. Below that central point, the justification follows subsumptive order and above that central point, the justification follows normal entailment order.
Contrast this with a network theory of justification. Here, no strict chains of belief would be found that followed any order involving entailment, either forward or reverse. Instead, each claim would contain its own set of links of justification to some other claims. Some claims would contain a greater number of such links to or from them than other claims, and some claims would contain stronger links of justification than other claims. It is analogous to the network database of HTML pages on the world wide web, where no pages are foundational. Every page is simply a node in a network, some nodes of which are more important than others purely because they receive more traffic. The equivalent of an important claim in a foundationalist theory is one where many different claims are justified in terms of it.
The clear advantage of a network database on the internet is that it makes for robust navigation via hypertext. If one of the pages is inaccessible, it can be bypassed quickly so that navigation can occur around it. The overall traffic of the network is therefore preserved. Analagously, a coherentist theory of justification can have the advantage that one's system of beliefs becomes much more robust. If one belief becomes questionable, the other beliefs can still hang together well because of their coherence with the other beliefs in the system. The entire system of belief does not collapse just because one or two claims in it are found to be faulty. Most beliefs in the network will rely on justification from more than one belief, and so they will not be jeopardised just because any one belief is thrown into question.
This all sounds very sophisticated and modern in theory, but it does not seem utilised in practice. Science and mathematics both rely on hierarchical justification. Even the assumptions of a paradigm form foundations for derived assumptions relating to the methods of hypothesis, experimentation, confirmation, refutation and so on. Allowing some of the assumptions to be justified in terms of the fruit they bear makes for some coherentism. But this is coherentism that preserves a rigorously hierarchical nature. Science has always been constructed like a building, with a foundation that allows for any edifice that might be erected upon it. Geometry provides a clear example of how sophisticated theorems are built up step by step from fundamental axioms.
In fact, it is hardly clear exactly what a network justification of science would even look like. But it certainly sounds chaotically unstructured on the face of it. The reason that science utilises a hierarchical system of justification is that it is easy to understand, debug and maintain. Coherentism allows for a greater flexibility in how chains of entailment can be justified, but to get rid of the chains would appear to be a grave mistake. Right now, if someone wants to question a particular scientific belief, the exact point of attack seems very clear. One can pinpoint exactly which link in the chain of entailment is weakest and strike there. If each claim in science had multiple points of justification, it would be much harder to arrive at conclusive refutation, which is an essential part of scientific progress, as Karl Popper recognised.
Therefore, coherentism may be admitted as a more fruitful explanation of the power of science. But it must be very clear what kind of coherentism is being implied. It is hardly an amorphous network of beliefs that are difficult to deny because of their multiple points of 'confirmation'. It would be very difficult to evolve such a system the way that science must evolve, because the very line between confirmation and refutation would become decidedly blurred. It would never be clear exactly when a hypothesis had been refuted because it might always link up to another claim somewhere else. If we keep a rigorous chain of entailment, however, science is far more capable of deciding between competing claims. Hence, a hierarchical coherentism must be supposed.
Comments