The power of "like"

1. Introduction

William Golding's novel The Inheritors tells the tale of a group of Neanderthals who are on the point of being supplanted by homo sapiens. The tale is told mostly from the Neaderthals' point of view.

In chapter 10 one of them, Lok, makes a great intellectual advance. Here is the paragraph:

"Lok discovered 'Like.' He had used likeness all his life without being aware of it. Fungi on a tree were ears, the word was the same but acquired a distinction by circumstances that could never apply to the sensitive things on the side of his head. Now, in a convulsion of the understanding Lok found himself using likeness as a tool as surely as ever he had used a stone to hack at sticks or meat. Likeness could grasp the white-faced hunters with a hand, could put them into the world where they were thinkable and not a random and unrelated irruption."

As the paragraph says, likenesses had been noted before. Here are four examples in the mouths of characters:

Chapter 3, Fa: "Today is like yesterday and tomorrow."

Chapter 3, Lok: "No plant like this grows near the fall!"

Chapter 5, Nil: "Now, is like when the fire flew away and ate up all the trees."

Chapter 5, Lok: "He is like a cat and he is not like a cat."

Lok's great advance is to become aware of a general tool of thought, likeness, that can be used when required. Its use makes possible a step forward into a world of properties as independent objects of discourse. If two items are recognised as different but like each other, we can think that there is some property in respect of which they are alike. (We shall use the neutral term "items" to cover physical objects, people, and abstract entities, including properties themselves.) Philosophers debate whether there are such things as properties, but unless we believe that the notion of a property is incoherent, we can happily and productively think and talk as if there are properties.

A ground of likeness might be immediately concrete, as in shape or colour. Or it might be less immediately concrete, as when two people are alike in character, meaning that now and in the future they will react in similar ways to circumstances. Or it might be more abstract, as when a hat and a formal title are alike in being markers of social status.

Once the existence of ways in which two items may be alike is recognised, thought can advance into a world of properties shared by items, can group items in easily remembered categories, and can build whole conceptual schemes. We may speculate that the grounding in concrete items and their immediately apparent properties was psychologically essential, that we could not have built a superstructure without such foundations even though that might have been logically possible. But once the step has been taken, properties and the relations between them can be explored in their own right.

We are here racing ahead of the Neanderthals as they appear in the novel. We do so in order to set the scene for our discussion.

2. "Like" and likeness

Is it the word "like" that confers power, or the ability deliberately to find likenesses?

In the paragraph quoted above, it is the word that is mentioned first. And this makes sense. It helps to have a word for a mental tool in order to think about it, or even to focus the mind to use it. (In the novel, the Neanderthals sometimes go beyond the limits of their spoken language by having pictures in their heads. That resource is however less likely to be productive with abstract notions than with physical objects.) But Lok goes on to use likeness, the notion, as a tool. It is after all notions and not words that can do intellectual work.

What is involved in the use of likeness as a tool? For it to be an advance in the way that Golding describes, the tool must be used deliberately, hence the need for the word "like" to call the tool into action. As noted, likeness was used unconsciously in earlier chapters. But there it was used as and when the idea of a particular likeness popped into someone's head. Once likeness is a recognised tool, it can be used in a planned way. The user can consciously look for likenesses between items, create categories in which to file items, and use categorisation as a way both to keep a grip on what has been seen and to think of items when they are no longer present. An item no longer present is recalled to have been in a particular category of things alike in some respect, and the thought of some things in that category can lead the mind to the thought of other things in the category. The mind no longer needs to search itself for "some specific thing not present". Instead it can search itself for "a specific human being not present" or "a specific red flower not present", using some example that comes easily to mind in order to start the search.

Although deliberate use does require having a notion of likeness, and a word to call the notion to mind, the notion need not be investigated. There is no need to reflect on what it is in general for items to be alike. But the mere happenstance use of the tool would not be enough to make the big advance that we have in mind here and that Golding attributes to Lok.

3. Types of likeness

3.1 Manifest property-sharing

3.1.1 Appropriate properties

Two items may be alike in that they share some property. We here have in mind manifest possession of a property, the attribution of which does not depend on metaphor or anything like that.

One indicator of likeness by virtue of manifest possession would be that if we listed the properties of each item separately, the property found on both lists by reference to which likeness was claimed would be one that it would have been appropriate to have put on those lists even if we had searched for properties of each item without having any comparison with the other item in mind. The intention here is to exclude properties that we might identify, or even define when they would not otherwise have been conceived at all, in order to establish likenesses between the two items. The world, and not an urge to find likenesses, should be in the driving seat.

Having said that, relational properties between each item and further items are not excluded. Two items might for example be alike in that they both originated in the same place. That place would be a third item, and each of the two items under consideration would have the relational property of having originated there.

3.1.2 Significance

It may be clear that two items share some property, allowing us to say that they are in that respect alike. But the property might be trivial. The currency of likeness would be devalued if likenesses were identified all the time on inconsequential grounds.

Whether a likeness is in respect of a sufficiently significant property may be determined by our purposes.

If our purposes are practical, we may pay particular attention to likenesses that, for example, would allow one item to be used as a substitute for the other or would indicate that dangers associated with one should also be associated with the other.

If our purposes are theoretical, we may select similarities on which to concentrate on the basis that they are the ones that reveal an underlying structure, such as a tree of evolution. We may deliberately ignore other similarities because they would obscure the picture.

3.2 Properties discovered on comparison

Two items may be regarded as alike despite not sharing properties that would naturally be attributed to each one separately without any thought of the other. The likeness may only emerge when we look at the items together and then spot a property that they share, a property that could have been noticed by reference to just one item but would probably not have been.

We first identify a whole set of properties that could naturally be attributed to the first item separately, giving between them a description of it (which may well be a partial description). Then we do the same for the second item. Finally we identify a commonality between the items by virtue of some further property, not on either list, that is possessed by both. That possession emerges when we review the descriptions of the items.

To take a basic example, we might describe a football and a tennis racket, include in the descriptions statements of how they were put to use, and then identify their common property of being inanimate objects used in sports.

To take a more abstract example, we might identify symmetries in each of two structures separately, and then identify the shared mathematical property of exhibiting symmetry groups.

At an intermediate level of abstraction, we might identify bad harvests at one stage in history and high taxes at another, then identify the common property of being drivers of social transformation.

Recognition of this kind of likeness requires more creativity than recognition of the sharing of manifest properties, because the common property will not be one that comes to mind without thought.

In addition, extra  thought will be needed to ensure that identifications of shared properties are helpful. It would usually be possible to contrive some shared property or other, just for the sake of doing so, but contrived properties are unlikely to be worth identifying. So whatever comes to mind must be subjected to criticism. In particular, would it resonate with other people, either people in general or people with relevant expertise?

If likenesses such as those discussed here are identified, they can be powerful drivers of discovery of more about the items. If we can compare items in one respect, maybe other features of one item will point to features of the other item that have hitherto gone unnoticed.

4. Properties as items in their own right

There is a base level of items that are not themselves properties. Those items may share certain properties which may naturally be attributed to them separately, and we may remark on their likenesses. This is the process we covered in section 3.1.

It is a more sophisticated process to do what we described in section 3.2 and identify a likeness on the basis of descriptions of two items, physical or otherwise, where neither description explicitly includes the property on the basis of which the likeness is asserted.

We can discern the following steps. First, look at the two sets of properties that constitute the descriptions side by side. Then identify a single property that each set separately implies, whether logically or in some looser sense of implication. Each set will imply a whole host of properties, but looking at the two sets side by side should both inspire our inferences and narrow the field to properties that are implied by both of the sets.

At this point, with implications between properties, we are dealing in properties rather than the items that have them. Any other item with the same set of properties as either of the two already considered would be like those two in the identified respect.

Moreover, the properties are items in their own right which may themselves be alike in exemplifying other properties. Thus the properties of being red and being green both exemplify the property of being coloured, and the properties of being difficult to produce and being widely desired both exemplify the property of having an influence on market price.

Talk about properties in their own right, rather than items that have them, could be inspired simply by remarking on likenesses in respect of properties we might naturally attribute to each item in isolation, the sort of thing we covered in section 3.1. But it is psychologically more likely when we have had to work out the properties to attribute.

A move to talking about properties in their own right opens up the opportunity to construct and study whole conceptual schemes. Then the data that the world supplies can be fitted into patterns, making recording, memory and analysis a great deal easier.

Another new opportunity is to think of occurrences of the properties independently of the existence of items that we have in fact encountered. There is scope to suppose new items, and to suppose novel combinations of properties.

There is also the opportunity, once we have the habit of thinking of properties in their own right, to invent variables that can take properties as values. Then we can quantify over properties, and develop second-order logic or informal talk along the same lines. This will allow us to say things like "All the properties of this item are also properties of that item". Second-order logic facilitates tremendous advances. But we shall not pursue the topic here, other than to note the psychological dependence on thinking of items as alike and then abstracting the properties in respect of which they are alike.

There is a further step that we shall also not pursue here, of reflecting on what we are doing when we use the mental tool of likeness. We have so far spoken of its use, and have noted that in order to make the great advance of deliberate use of the tool, we need to have some notion of likeness. But we could also consider the notion of likeness in its own right. Such consideration could extend to considering the ways in which different likenesses are similar or different, for example as strong or weak, as physical or abstract, or, within the physical, as based on appearance to human senses or on underlying mathematical structures.

5. Tags and hierarchies

Once properties are considered as items in their own right, it becomes possible to construct conceptual schemes that set out maps of properties. Different ways to do so are analogous to some of the ways we now use to keep control of emails and other files on our computers.

Reliance on a notion of likeness would encourage a mental equivalent of the tagging that is available in some file management systems. Any item, whether one at the base level or a property, could be tagged with the names of one or several properties, without the need to comply with any hierarchy or other systematic pattern. When two items struck us as alike in some respect, the tag for the appropriate property would be used to record the likeness. And if there were no existing tag, an appropriate property would be defined and its name used as a tag. Even before we got on to conceptual schemes, tagging would yield the considerable benefit of grouping items in categories that were easy to remember. It would also facilitate thought about items not present and items that were purely hypothetical.

Another approach that could be used, instead of or alongside tagging, would be the construction of hierarchies of properties, analogous to a tree of folders and sub-folders. At the top would be the most generally applicable property within the given field (physics, or history, or motor mechanics, or hunting, or whatever it might be). Beneath it would be properties that would be applicable to narrower ranges of items, and so on. A base-level item would be considered to have all the properties above any property it was considered to have, and a property would be considered to exemplify all the properties above it.

Hierarchies would not spring so directly from the notion of likeness as tagging would spring. And a hierarchy could be limiting, as when two items in different branches were thought to belong together in some respect. That could however be addressed either by adding a tagging system or by having different hierarchies of the same properties, constructed on different principles or with different highest properties, for different purposes.

Differences between the two approaches also emerge when we consider the development of a deep understanding of the world. Suppose that we managed to formulate a hierarchy that was found to categorise items smoothly without leaving many items that would be hard to place in the hierarchy or that could legitimately be placed in several competing locations. Such an exercise would have a good prospect of bringing large-scale facts about the structure of the world to light. A hierarchy is not itself a theory, but the principles on which it is constructed may amount to a theory. And the theory would derive credibility from the fact that the principles allowed the smooth construction of a hierarchy that gave a useful map of the world. The use of tags may also lead to a deep understanding, but tagging can be based on casually noticed and imprecisely defined likenesses. So it is less likely to yield a deep understanding than the struggle to formulate a hierarchy that works.

While tagging may be too easy to be a reliable path to deep understanding, it does have the advantage of flexibility. New items, whether base-level or properties, can be brought into a tagging system without much thought. Their relationships to other items can be sorted out later. Making connections in a way not constrained by an existing understanding of a hierarchy can itself be a route to new ideas. And tagging is the easiest way to take forward the use of likeness as a tool. If two items are alike we can give them a shared tag, either an existing one or a newly created one. Then there can be a stage of noticing which other items have or should have the same tag, or a stage of reviewing everything and noting when particular combinations of several tags recur in relation to several different items.

6. Conclusion

We have loosely and speculatively sketched the role of spotting and recording likenesses in the intellectual development of humanity.

The tool of likeness is however not enough. The development of a powerful understanding of the world, whether in the natural sciences, the social sciences or the humanities, requires something considerably more sophisticated than "this is like that", although much can be done by forcing oneself to say "this is like that in such-and-such a respect", while demanding a pretty precise specification of the respect. 

Even that is still far from enough. Theories do not consist solely of catalogues of likenesses and explicit or implicit principles on which the identification of likenesses is based. They must identify principles on the basis of which the relevant aspects of the world work. In the natural sciences those principles should explain the likenesses that can be observed. In other disciplines they should at least make observed patterns of likenesses less surprising than alternative patterns that we could imagine.

Nonetheless, discovery of the power of "like", whether first made by Neanderthals or by homo sapiens, was a vital early step.