The Difficulties in Symbol Grounding Problem and the Direction for Solving It
![](/bundles/mdpisciprofileslink/img/unknown-user.png)
Round 1
Reviewer 1 Report (Previous Reviewer 1)
Terrence Deacon whose approach to the symbol grounding problem you ignore at your own expense. I think the entire cog sci methodology you employ in this paper's research is fundamentally flawed. I disagree wholeheartedly with your conclusions. And I recognize that you are well within the range of normal science approaches to these questions.
I think of the hard problems as "made harder problems" by means of your core assumptions. If you never explain the origin of effort of work that works to keep a being working within its work setting (That is, beings making functional interpretive effort) and you start with computational models from which to derive qualia, sure the problem is hard but also crypto-cartesian.
Totally unimpressed with the argument you make though you make it with great clarity and excellent reference to the existing literature.
If you're serious about your research questions, I believe you'd find something useful in Deacon's approach.
Author Response
Dear reviewer,
Thanks for your comment. We have to admit that we are not professional in Deacon’s theory. Following your suggestions, we have read some materials about Deacon, especially his theory of symbols in the book the symbolic species. And we also found an article discussing how to solve symbol grounding problems based on his theory: Raczaszek-Leonardi, J. and Deacon, T. (2018) Ungrounding symbols in language development: implications for modeling emergent symbolic communication in artificial systems. 2018 Joint IEEE 8th International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob), pp. 16-20. https://doi: 10.1109/DEVLRN.2018.8761016
From the limited understanding of his theory these days, we recognize that his theory is comprehensive and profound. And we cited and commented on his theory in part 2.2 in order that our paper is more informative. Though his theory is important, from the view of Harnad and others who specialized at SGP, the intentionality problem related to consciousness is still the main obstacle to solving it. we didn’t see many discussions on Deacon’s theory. It is possible to give a unique answer to SGP by Deacon’s theory, but it needs a quite long time to form and is not the purpose of this paper.
It is natural to discuss SGP from the computational model because SGP itself is formed in this area. It is hard to see how to talk about SGP in other research fields. We humans do not have SGP because the symbols we use obviously have meanings. Deacon focused on how language came about in human beings but that’s another question. We are glad to know more about Deacon, but in this paper, time is not allowed. I would take it as one aspect of my future research.
Reviewer 2 Report (Previous Reviewer 3)
The author[s] should correct grammar and syntax errors and shoud discuss with personal critical observations the theories they refer to.
Author Response
Dear reviewer,
Thanks for your comment. We have checked the grammar of this paper again and added some new comments and references. We have given comments for most theories in this paper. We are pleased to hear more of your advice if you give more details on where you think we did not give a good discussion.
Reviewer 3 Report (New Reviewer)
Please see the attachment
Comments for author File: Comments.pdf
Author Response
Dear Reviewer,
Thank you very much for reviewing. We made some revisions according to your suggestion. Please see the attachment.
Thank you again.
Li Jianhui and Mao Haohao
Author Response File: Author Response.pdf
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
This is a fine article by today's academic standards in the field. I consider it a typical example of abstract modeling, the default in philosophy.
I distinguish between formal falsifiability (internal consistency), empirical falsifiability (predictive utility) and explanatory falsifiability (accurate representation for how the phenomena in question actually manifests in the real world.
This paper deals with the first two, but like most of this kind of writing, ignores the latter, which is singularly disappointing given that it's a paper on how a semiotic relationship is actually grounded in the real world.
The authors attend to symbols, feeling and consciousness, never once noticing that feeling and consciousness are themselves grounded in vegetative sentience, effort, the struggle for existence against the second law of thermodynamics that is common to all organisms. It's as if humans are determinate computational machines that experience qualia and consciousness. This is not a fault of the authors, who do excellent work here. Rather it's a problem with the entire disciplines of philosophy of mind and cognitive studies.
I recommend this paper:
https://www.researchgate.net/publication/334487464_Ungrounding_symbols_in_language_development_implications_for_modeling_emergent_symbolic_communication_in_artificial_systems
Reviewer 2 Report
The paper under review is a contribution to the debate over the Symbol Grounding Problem (SGP). The main problem is that it does not offer any novel insights and remains a fairly superficially written literature review, with some minor mishaps and argumentative obscurities. I cannot really see how this paper could be revised to make it publishable, mostly because it does not bring any new interesting points.
It's fairly obvious that Searle connected consciousness and meaning, and that some agree with this, and some don't. It's well-known that Dennett criticized the notion of original intentionality. So what's new here?
There are also serious problems with writing, starting from the English, but mostly related to the impressionistic style of argumentation, which remains obscure and gappy at multiple places. There are a number of lengthy citations that make this paper look like a series of notes one made when reading. But they should be succinctly summarized, as they seem to bring not so much to the paper itself.
There are argumentative obscurities in particular related to the charge that various authors have contradictory views, but it is not explicitly stated which claims contradict which. I cannot see any contradiction anywhere in the passages cited by the author. It's extremely uncharitable and the such writing seems implausible. In particular, line 287-291, there is no contradiction between the statement 'artificial intelligence is programmed by humans' and 'computer programs have intrinsic intentionality'. If there is, please show where! The contradiction only occurs if you add some additional premises, such as 'if X is programmed, then X has no intrinsic intentionality'. But this is begging the question against Harnad, and not certainly his premise. Moreover, human DNA may be seen as a kind of program, and we have certainly not programmed ourselves, while Searle believes we exhibit intrinsic intentionality. The author seems to make an argument that is similar to Lady Lovelace's here, but this argument is known to be fallacious, or at least controversial. Anyway, there is no contradiction in Harnad, even if the author repeats this ad nauseam (line 300, line 303).
All in all, the author seems to be oblivious to the fact that the SGP addresses the same problem that has been already taken by the teleosemantic tradition in naturalizing intentionality, and there's a fairly big consensus that Millikan's approach was successful (in spite of what Mendelovici says, but her position is in the minority). Read up on mental representation; Nick Shea's recent book may be very helpful.
Minor comments:
Line 14 and 355: 'Chinese Room argument': there is no proper argument there, as Dennett noticed, there's a thought experiment but neither the premises nor the conclusion are explicitly stated in the original paper.
L. 15: John Searle barely needs an introduction, and calling him 'a leading American professor of the philosophy of mind' is not accurate because he was denied the title of the Professor Emeritus at Berkeley because of his long history of sexual abuse. He is thus no longer a professor.
L17 'proceed': you mean 'process'
L18 The sentence starting with 'While' should be part of the previous one (sentence fragment)
L19 'can't': it should be 'couldn't' (tenses!)
L20 'of understanding': remove 'of'
L23 'proceeded': you mean 'processed'?
L23 'can understanding': it should be 'can understand'
L28: 'because' should not be at the end of the sentence
L55-57: this sentence is redundant, it was stated already before on lines 33-36
L73 'syntactics': the standard term is 'syntax'
L79 "character 'apple'": in English, this is a multicharacter word, maybe the author means some ideographic script but it makes little sense here
L110: Semiotics stems from Peirce but not only; there is also a large tradition that was started by de Saussure. All in all, it's not a single theory but a large research tradition that is inaccurately presented in the paper.
L.117-120 and lines 189-191: Semiotics DOES NOT pressupose that all symbols mean anything. This is absurd. Obviously, there might be vacuous terms, and Peirce's writings are full of examples of deficient signs. Applying semiotic categories basically changes nothing in the SGP. The author is completely mistaken.
L243: I don't see why the author cited Bielecka 2016 but not her 2015 paper on Floridi and Taddeo and why they failed to solve the SGP. This would be relevant here:
Bielecka, K. (2015). Why Taddeo and Floridi did not solve the symbol grounding problem. Journal of Experimental & Theoretical Artificial Intelligence, 27(1), 79–93. https://doi.org/10.1080/0952813X.2014.940138L247: it is not true that 'the notion that the problem of consciousness underlies the symbol grounding problem has emerged gradually' Basically anyone in the debate knew that Searle connected consciousness and intentionality; it's just many people don't agree with it (line 251: Harnad was definitely not the first one to say that)
L266: 'mentalis': it should be 'mental is'
L324: what is 'T3-capacity'? explain this terminology if it's needed, or drop the citation
line 327 'placed meaning and mind on an equal level' - unclear, do you mean 'conflated meaning and mind'? what level?
L 357-358: this is obscure, cannot understand what the author means
L365 unclear why some peoples' are names are given with the first name and surname, and some not; and definitely not all of them are computer science professors
L384 and elsewhere: I don't understand what the author means by 'slide into dichotomy' or 'slip into dichotomy' (line 406), and it's unclear that any dichotomy is present in Davidsson based on what is stated in the paper; I cannot see what's dangerous about 'slipping into the dichotomy' (line 397). This should be made explicit!
Note that the hard and easy problem of consciousness is not a dichotomy at all. For all we know, there could be a spectrum from harder to easier problems, so a binary division is not really defensible. Moreover, it's not clear that there is a hard problem of consciousness anyway (it seems to be believed to exist by dualists and only some physicalists). I would certainly think it's a pseudoproblem in Carnap's sense of the term.
L422-428: very obscure paragraph
L430 one should use just one spelling of 'Müller' (not 'Mueller' and 'Müller' in the same paper)
L510: Dretske is usually referred to as 'Fred', not 'Frederick'
Reviewer 3 Report
The aim of the paper is to investigate the so called Symbol Grounding Problem, in order to open a track apt to solve the question of how – and, I would add, “if”- a symbolic system can acquire semantic properties. The author(s) correctly traces the origin of the problem back to the argument of the Chinese room of John Searle and suggests to divide it in hard and easy problem, as Chalmers proposed to do in reference to consciousness. The author(s) quite surprisingly doesn’t mention Chalmers and, even more surprisingly, doesn’t explain for which reason the SGP would benefit if distinguished into two problems, an easy and a hard one.
The author(s) mainly refers to the work of Taddeo and Floridi, and secondarily to Harnad, Steels and Vogt, Brooks, Mueller, whose writings are merely summarized, without critical discussion or evaluation. Externalism, also called tracking theory, is introduced as one of the two candidates to solve the SGP, but it is quite obscure the reason why externalism could help solving the SGP, since “externalism” is vaguely defined as the theory that considers mental states to be influenced by external world. The problem of SGP is not if mental states are only internal or are influenced by the external world, but if and how sinctatic symbols can acquire semantic properties, i.e. meaning.
In chapter 4 the author(s) goes back to Searle, and exposes the distinction between intrinsic and derivative intentionality proposed by Searle in the context of philosophy of mind problems. The reason to mention the distinction remains unclear, since it is not exploited in the following, and the conclusion that phenomenal intentionality theory is the other good candidate to solve the SGP, is stated without giving any explanation to support such proposal.
The author(s) shows a good capacity to sum up the selected papers, also difficult ones, but this positive attitude is insufficient to recommend this paper for publication, as it doesn’t meet the standards of a good level philosophical paper since it exhibits poor critical thinking, little or none discussion attitude, no conceptual debate, no personal proposals. Moreover English language is not always clear and is often not correct.