Computers don't learn things they remember them, like a child memorizing a poem or a math formula or a rude word in a foreign language. I like your distinction about apprehending patterns, I think that explains the chin room, the computer apprehends rather than comprehends.

Humans aren't capable of grasping the essential nature of objects either, and that ties into your argument, perhaps rephrasing it a little.

One may look at an object or phenomenon, gravity for example, and say things about its behavior, its properties and qualities. It is then by these factors one determines one is observing the workings of gravity. But what is gravity? One can give a fairly accurate model for gravity, but it remains a model. A sound man produces to describe the observation of a certain phenomenon. It is not true understanding of the essential nature of what man calls gravity. So in a sense, a human being is also simulating understanding, without ever achieving true understanding, just like the computer in the Chinese Room argument.

I think comprehension sans consciousness isn't an unattainable goal for computers.

Since comprehension is about meaning I laid out the basics of it in my OP. There are a variety of definition types and methods.

Definition types:
1. Lexical definition: this uses analytic definitions; genus and species
2. Theoretical definition: used in philosophy, science and math as part of a theory
3. Stipulative definiton: new or old terms given a new meaning.
4.. Precising definition: to get rid of vagueness
5.. Persuasive definition: emotionally charged to influence readers

Methods of defining:
1. Ostensive: pointing to objects the term applies to
2. Synonymous definition: using other words with the same meaning
3. Analytics definition: this is the standard method through identifying a larger class (genus) and then specifying a smaller subgroup (species)

The subtext to all the types and methods is recognition of pattern in the properties or association of given term with others.

I differentiate pattern from association because there's a significant difference between synonymous definitions and analytic definitions. In the first we can understand by saying a term is equivalent to another and another e.g. water=H20=aqua etc. There's no need to see pattern in this case. In the second it is essential for us to see pattern in what is given in the genus class and the species class. However, synonymous definitions eventully must be given a seed analytic definition if one is to comprehend them.

There really is no comprehension if there is no pattern recognition.

So, in my humble opinion, comprehension is all about seeing patterns and these patterns are associated through the structure in these patterns. The more phenomena we link through their patterns, the more the comprehension. If I were to draw an analogy (pattern recognition) comprehension is like a house. Each brick is a pattern and the end result of all those bricks interlocking with eachother forms the house/comprehension. I agree the house has more qualities than a brick but the point is the bricks and their interactions constitute the house.

The Chinese Room Argument is supposed to demonstrate that computers can't comprehend like humans do.

I agree that is true in the present. Yet, I ''feel'' that if comprehension is simple pattern recognition then it can be done by computers. For example a computer can be given the input 3 dogs, 3 cats, 3 apples, 3 cars and programmed to look for similarities/patterns in the input. It would immediately recognize 3 to be the pattern and then comprehend the meaning of 3-ness.

My:
Strong conclusion: computers can be made to comprehend
Weak conclusion: Human comprehension isn't as special as is portrayed in the Chinese Room Argument

I agree that a lot of our learning is also just remembering symbols that we don't really understand in themselves, only in connection with other symbols. I believe that a weak level of comprehension is possible for AI, indeed we are almost there now.

The scenario that we will create a strong AI that will be smarter than humans though I think is a bit far fetched. AI could be taught to 'do philosophy' but not to 'make philosophy'.

Meaning or definition occur to us in two forms:
1. Intensional meaning wherein we list the properties of a class (genus) and then distinguish its members using other propeties (species)

E.g. A fish is an aquatic animal

2.Extensional meaning in which we list the objects that have the properties given in the intensional meaning.

E.g. Trout, salmon, sturgeon are fish — TheMadFool

Meaning isn't the same thing as a definition, and meaning isn't captured by saying something like "A fish is an aquatic animal." Meaning is what's going on mentally behind saying "A fish is an aquatic animal." It's the mental act of (non-arbitrarily) making that sort of association. That mental, associative act can't be captured in a text string or a set of sounds, and a text string or a set of sounds can't somehow do the relevant mental, associative act.

AI could be taught to 'do philosophy' but not to 'make philosophy' — Jamesk

Are you referring to creativity?

A book I read states that creativity consists of combing old ideas into new ones. For instance a phone = electricity + sound and a mobile phone = phone + radio. This type of creativity seems possible with computers.

Art seems more difficult. I've heard of computers composing music but not paintings.

Meaning isn't the same thing as a definition — Terrapin Station

I was just thinking about that. Sentences convey a different level of meaning but what is of note is that a sentence consists of defined words. For instance the sentence ''water flows'' consists of the definitions ''water'' and ''flow'' and each of the two is a pattern which a computer can be programmed to recognize. If a hypothetical computer is given the ability to recognize paterns anf then fed the inputs ''water flows'', ''oil flows'' and ''air flows'', it should be able to comprehend the meaning of both ''fluid'' and ''flow''.

Are you referring to creativity? — TheMadFool

Perhaps a form of 'original creativity', Perhaps a 'conscience'. The property of being Human is hard to define but cannot be ignored.

Definitions are for example the text strings you find in dictionaries, or the string of spoken phonemes re a verbally given definition.

Meaning is a mental act of associations that we perform in our heads.

mental act of associations — Terrapin Station

Coming to associations:

There are different types of association:

1. Non-pattern association: You're having lunch in at a Baker street restaurant. Suddenly there's a shooting outside and some people are injured. Next time you see/hear the word ''Baker'' you'll immediately remember gun or death or whatever about the incident that got your attention.

There is no pattern in this association. It is just a coincidence of events that you've associated and it may have personal meaning but there's no causal connection to it.

2. Pattern association: A child sees a tree and its characteristic branching structure. Later the same child studies phylogeny (evolutionary relationship of living things) and notices the same branching structure. It's called a phylogentic tree. Here the association is by pattern.

It may be easier to teach computers type 2 associations than 1 because the latter requires a computer to have sensory input at a scale unimaginable now.