At around 8,500 words, the piece is significantly too long for a typical Substack post. I strongly recommend breaking it into a series of shorter articles, each focused on a distinct theme (e.g. categorisation, memory, language acquisition). Readers are more likely to stay engaged with posts under 2,000 words. I quickly disconnected and it will take quite a while to read and make sense of your thesis.
The central idea—that intelligence operates via comparison and that this underpins language—is intriguing, but the argument lacks clear logical scaffolding. Key claims (e.g. “differences are faster than similarities”) are asserted rather than demonstrated, and the transitions between topics can feel abrupt.
As a linguist, I find you assertion on Grice very misguided because you do not define the scope.
It's about contributing to a conversation in a way that is oriented toward mutual understanding, assuming a shared goal of communicative success. The four maxims—Quantity, Quality, Relevance, and Manner—are guidelines that rational interlocutors are presumed to follow unless there's good reason not to.
Example: Saying “Some of the guests arrived” when you know all did violates Quantity, but might implicate that not all were welcome.
Grice himself acknowledged that violations or floutings of these maxims are not only common but essential to communication, as they generate implicatures. So the principle was never absolute.
It is really good to see that you cite several important thinkers (e.g. Gärdenfors, Goldstone, Tomasello), but references are occasionally vague or used without critical engagement. The piece would benefit from tighter integration of sources into the argument and a consistent referencing style.
I think this has has real potential if reframed as a structured series. I wish better skilled people than me wold engage with your piece as well. I would suggest you redraft starting with s short introductory post outlining the overarching thesis and then with follow-up posts on:
- categorisation;
- memory;
- object recognition;
- language acquisition;
- reference and meaning.
You more concret examples if possible, it helps with bastract claims.
Sorr this is very short and doe not address the sentral thesis. As I said above, a very long piece to read!
I will try and focus on the thesis at a later stage.
Best regards
More concrete examples, diagrams, or scenarios to support abstract claims
"Key claims (e.g. “differences are faster than similarities”) are asserted rather than demonstrated" - this sounds strange given the content of the paragraph where that claim is made.
I will appreciate your expanded comments. Sorry for making it so long. I wanted to connect all the dots in one piece to minimize the need to go back and forth between different posts.
I think the issue for me is that dots you are connecting are very clear in your mind and make sense to you. But from the outside, one has to recognise the dots, work out the assemblage and look for connections and then relate all of this to their own knowledge and way of approaching knowledge. I think smaller works better, but it may be just me.
Regarding 'differences': your paragraph does illustrate the idea that differences can be used efficiently in classification (as in 20 Questions), but the claim that “differences are faster than similarities” is presented as self-evident, but not *empirically demonstrated*.
Your argument is *theoretical*, relying on an analogy rather than experimental evidence, unlike *similarity-based models* (Rosch’s Prototype Theory and Nosofsky’s Generalised Context Model) which have been *empirically tested*, showing that similarity effects influence categorisation speed and typicality judgements.
I could also be argued that the assumption that a difference-based tree search is faster does not account for cognitive realities: categorisation in humans is often *graded* (Rosch 1978), *fuzzy* (Rosch & Mervis 1975; Barsalou 1987), and *context-sensitive* (Barsalou 1987; Murphy 2002), not binary. Dual-process theories also show that reasoning may alternate between intuitive and logical modes, depending on context and task demands (James, Stanovich, West, Kahneman).
I do not believe that the efficiency of “differences” over “similarities” is established by the example you give alone. I like the hypothesis, but I think it would need experimental support to move from illustration to evidence.
Hi!
Hope you are doing well.
At around 8,500 words, the piece is significantly too long for a typical Substack post. I strongly recommend breaking it into a series of shorter articles, each focused on a distinct theme (e.g. categorisation, memory, language acquisition). Readers are more likely to stay engaged with posts under 2,000 words. I quickly disconnected and it will take quite a while to read and make sense of your thesis.
The central idea—that intelligence operates via comparison and that this underpins language—is intriguing, but the argument lacks clear logical scaffolding. Key claims (e.g. “differences are faster than similarities”) are asserted rather than demonstrated, and the transitions between topics can feel abrupt.
As a linguist, I find you assertion on Grice very misguided because you do not define the scope.
It's about contributing to a conversation in a way that is oriented toward mutual understanding, assuming a shared goal of communicative success. The four maxims—Quantity, Quality, Relevance, and Manner—are guidelines that rational interlocutors are presumed to follow unless there's good reason not to.
Example: Saying “Some of the guests arrived” when you know all did violates Quantity, but might implicate that not all were welcome.
Grice himself acknowledged that violations or floutings of these maxims are not only common but essential to communication, as they generate implicatures. So the principle was never absolute.
It is really good to see that you cite several important thinkers (e.g. Gärdenfors, Goldstone, Tomasello), but references are occasionally vague or used without critical engagement. The piece would benefit from tighter integration of sources into the argument and a consistent referencing style.
I think this has has real potential if reframed as a structured series. I wish better skilled people than me wold engage with your piece as well. I would suggest you redraft starting with s short introductory post outlining the overarching thesis and then with follow-up posts on:
- categorisation;
- memory;
- object recognition;
- language acquisition;
- reference and meaning.
You more concret examples if possible, it helps with bastract claims.
Sorr this is very short and doe not address the sentral thesis. As I said above, a very long piece to read!
I will try and focus on the thesis at a later stage.
Best regards
More concrete examples, diagrams, or scenarios to support abstract claims
"Key claims (e.g. “differences are faster than similarities”) are asserted rather than demonstrated" - this sounds strange given the content of the paragraph where that claim is made.
I will appreciate your expanded comments. Sorry for making it so long. I wanted to connect all the dots in one piece to minimize the need to go back and forth between different posts.
I think the issue for me is that dots you are connecting are very clear in your mind and make sense to you. But from the outside, one has to recognise the dots, work out the assemblage and look for connections and then relate all of this to their own knowledge and way of approaching knowledge. I think smaller works better, but it may be just me.
Regarding 'differences': your paragraph does illustrate the idea that differences can be used efficiently in classification (as in 20 Questions), but the claim that “differences are faster than similarities” is presented as self-evident, but not *empirically demonstrated*.
Your argument is *theoretical*, relying on an analogy rather than experimental evidence, unlike *similarity-based models* (Rosch’s Prototype Theory and Nosofsky’s Generalised Context Model) which have been *empirically tested*, showing that similarity effects influence categorisation speed and typicality judgements.
I could also be argued that the assumption that a difference-based tree search is faster does not account for cognitive realities: categorisation in humans is often *graded* (Rosch 1978), *fuzzy* (Rosch & Mervis 1975; Barsalou 1987), and *context-sensitive* (Barsalou 1987; Murphy 2002), not binary. Dual-process theories also show that reasoning may alternate between intuitive and logical modes, depending on context and task demands (James, Stanovich, West, Kahneman).
I do not believe that the efficiency of “differences” over “similarities” is established by the example you give alone. I like the hypothesis, but I think it would need experimental support to move from illustration to evidence.
It makes sense. I will think about experiments.