As we know, Peirce's writings have inspired many new theories and discoveries for well over a century. But we must always distinguish his exact words from anybody else's interpretations and extensions.
For interpretants, I believe that an article Jay Zeman wrote in 1977 is still one of the best: "Peirce's Theory of Signs", which has 70 occurrences of the word 'interpretant',
Since CP is the only source he cites, he avoids the dubious late extensions that Short and others have criticized. It provides a good foundation for readers to distinguish Peirce's earlier, more limited definitions from later extensions that Peirce himself failed to define clearly and precisely.
I have a copy of that article on my website, and I checked Google to find a more official site. But the search pointed back to my own website: https://www.jfsowa.com/ikl/Zeman77.pdf
After Zeman died, I downloaded a complete copy of his entire website. I believe that it should be preserved somewhere more official. If anybody knows of a more official place to store it, I would be happy to give them access to the whole thing.
John
Edwina, Gary, Jon, List,
As Peirce frequently pointed out, he had a solid understanding of all the methods of reasoning from the ancient Greeks to the medieval Scholastics to the methods from the Renaissance to the early 20th C.
In general, the "proper way" depends very much on the theorem proving algorithms. Peirce did not invent the method of transforming a sentence to make 'is' the verb that connects subject and predicate. That method was invented by Aristotle and systematized by Boethius. It was widely used in 19th c textbooks,
which were the ones that everybody, including Peirce, had studied .
For the differences between Plato and Aristotle, see slides 13 to 24 of https://jfsowa.com/talks/patolog1.pdf . For Aristotle's syllogisms and the methods for transforming sentences to make 'is' the main verb, see slides 25 to 32. The remainder of patolog1 discusses other patterns of logic from the middle ages to modern times, including those by Frege and Peirce. For more about modern methods, including Peirce's influence on them, see patolog2, 3, 4, and 5.
Fundamental principle: The methods of transforming formal logics are reversible. Anything represented in one format can be translated back and forth without loss of information. But transformations to and from natural languages and other formats -- formal, informal, linguistic, or graphic -- can lose information (or even worse DISTORT or CORRUPT Information).
When Peirce wrote anything on logic, he assumed that his readers were familiar with the kind of material summarized in patolog1.pdf. It's helpful to read that in order to distinguish Peirce's innovations from his sources and his assumptions about his readers. Modern methods of reasoning often transform the logic to different formats to adapt them to various algorithms. Those transformations are reversible (provided that both formats have equal expressibility).
Re dicisigns: Stjernfelt emphasized the trichotomy of rheme, dicisign, and argument because he was developing methods of reasoning with sources in natural languages.
But Peirce moved to the more general triad of seme, pheme, and delome, which allow diagrams and images as representations. He made that switch in 1904-5 when he was mapping images in the phaneron to existential graphs. Note that he continued to use the term 'phemic sheet' up to the end. But he never used the word 'dicisign' after he introduced the word 'pheme'. That is an indication of the way his theories were developing. And I believe that his correspondence with Lady Welby had a strong influence on that development.
John
_______________________________________
From: "Edwina Taborsky" <taborsky(a)primus.ca>
Gary, list
Thanks - that book however, is from ten years ago. My point is that current research in information dynamics in the ’natural realms’ - which, very often, doesn’t use Peircean terms but is obviously working within the same analytic framework of morphological formation, information generation, transmission and transformation, and the nature of ‘objective idealism’ [ the integration of matter and mind] …is extensive. I’d say that these are all analyses well within the notion of the dicisign- ie, the concept that information generation, processing etc is not dependent on language or even consciousness but is a basic process in the biological and physico-chemical realms. ..operative within sensate rather than symbolic networking. And - I’d say that these fit the definition of a dicisign propositional interaction, where meanings [Interpretants] are in direct or factual connection to the object. The problem is - as noted - this research doesn’t use Peircean terminology!
Edwina
On Feb 11, 2024, at 11:39 AM, Gary Richmond <gary.richmond(a)gmail.com> wrote:
Edwina, List,
ET: I’m a bit surprised by your request - since surely you are aware of the focus in science of Peircean principles in the biological, physic-chemical and artificial worlds.
Yes, I'm well aware of that focus and have read extensively in the literature. Thank you, though, for providing some recent examples, none of which I knew.
My question, however, specifically pointed to Stjernfelt's work on dicisigns. I wrote: "I've been wondering how this move of Peirce [throwing everything possible into the subject] might figure, if it plays a part at all, in consideration of what Stjernfelt called 'Natural Propositions'," the dicisign being a generalization of the proposition. Helmut Pape called Stjernfelt's book "an inter- and transdisciplinary study that discusses and criticizes theories and uses examples coming from psychology, biology, anthropology, neuroscience, biosemiotics etc."
So, mine was not a general question as to the influence of Peirce in fields other than logic -- there is no question of that -- but of this particular matter which Jon discusses at length and which may nor may not extend beyond logic as we generally think of it.
Btw, here is another book I'd highly recommend for discussions of Peirce's influence in biosemiotics in particular, edited by two fine Peirce scholars.
Peirce and Biosemiotics: A Guess at the Riddle of Life Vinicius Romanini (Editor), Eliseo Fernández (Editor)
The table of contents suggests the depth of the scholarship both by distinguished Peirce scholars and equally distinguished biosemioticians.
Table of Contents
Introduction; V. Romanini, E. Fernández.- 1. The Intelligible Universe; N. Houser.- 2. The Continuity of Life: On Peirce’s Objective Idealism; I.A. Ibri.- 3. Peircian Semiotic Indeterminacy and Its Relevance for Biosemiotics; R. Lane.- 4. Peircean Habits, Broken Symmetries, and Biosemiotics; E. Fernández.- 5. Semeiotic Causation and the Breath of Life; M. Hulswit, V. Romanini.- 6. The Ineffable, the Individual, and the Intelligible: Peircean Reflections on the Innate Ingenuity of the Human Animal; V. Colapietro.- 7. Instinct and Abduction in the Peircean Informational Perspective: Contributions to Biosemiotics; L.F. Barbosa da Silveira, M.E. Quilici Gonzalez.- 8. The Life of Symbols and Other Legisigns: More than a mere Metaphor?; W. Nöth.- 9. Signs without Minds; J. Collier.- 10. Dicent Symbols and Proto-propositions in Biological Mimicry; J. Queiroz.- 11. Semeiosis as a Living Process; V. Romanini.
Best,
Gary Richmond
On Sun, Feb 11, 2024 at 10:34 AM Edwina Taborsky <taborsky(a)primus.ca> wrote:
Gary R,list
I’m a bit surprised by your request - since surely you are aware of the focus in science of Peircean principles in the biological, physic-chemical and artificial worlds.
A few quick examples:
Homeostasis and Information Processing: Eduardo Mizraji. Biosystems 2024 February
"Teleonomic objects are purpose-oriented entities lacking a creator (Igamberdiev, 2023).
"What is the nature of the various types of information that different living beings, at different scales of their organization, use to inform their regulatory systems? This information is the result of the simultaneous existence of patterns and receptors capable of detecting these patterns. In this context, information is an emergent property of the interaction of two categories of entities, patterns and receptors.
This duo of conditions, thermodynamic openness and the need for homeostatic regulatory systems, are inherent to all forms of life that we know."
———————
Computation in Biological Systems as a quantum mechanical simulation .
Ron Cottam. Biosystems April 2022.
This article actually references Peirce.
——————————————————
The Information Continuum Model of Evolution BioSystems November 2021 R.Skern Mauritzen
---------------------------------------------------------------------------------
The Biosystems journal is a major site for the examination of informational processing within the natural world. That is - the focus is on information processes - and this, after all, is what Peircean semiotics is all about. [And there are other journals tha examine information processing]...
As I’ve kept saying, you don’t have to use Peircean terms to examine the same outlines Peirce was writing about …Unfortunately, the tendency of many Peircean scholars, to insist, almost with a religious and even cult-like fervour, on the use of exact and specific Peircean terms, obscures the fact that the same objective processes are being outlined in many papers - but- using different terms.
It is no secret that the difficulty of both accessing Peircean texts AND the obscure terminology [AND, I’ll add, the isolationism of the Peircean scholarly set] has hindered the widespread use of Peircean theories. But -once past these Walls - it is clear that the Peircean analysis is a fundamental outline of both the natural and human worlds of information processing.
Edwina
On Feb 11, 2024, at 12:23 AM, Gary Richmond <gary.richmond(a)gmail.com> wrote:
Jon, List,
Thanks for these additional comments and examples as they further clarify Peirce's bold logical move. As you noted:
JFS: Throwing everything possible into the subject recognizes the indexical nature of most words--functioning much like proper names, since one must already be acquainted with their objects in order to understand them--and leaves only the pure/continuous predicate as the iconic part of the proposition.
Since you mentioned Frederik Stjernfelt, I've been wondering how this move of Peirce might figure in consideration of what Stjernfelt called "Natural Propositions." In his extraordinary book, Natural Propositions: The Actuality of Peirce's Doctrine of Dicisigns, Stjernfelt takes Peirce at his word, so to speak, and generalizes the meaning of proposition well beyond the logical-linguistic into the natural world, that which biosemioticians naturally have a particular interested in. I don't know what lasting impact -- if any -- his book has had in that community since, as far as I can tell, it has been somewhat resistant to Peircean thinking. This antipathy was suggested to me when I attended a Biosemiotics Gathering at Roosevelt University in NYC some years ago (I was asked by Vinicius Romanini to read a paper he himself could not deliver since officials refused to allow him to board a plane to the USA because he'd brought his Italian, rather than his Brazilian, passport to the airport) as the several Peircean-inspired biosemioticians present seemed to be contradicted at every turn.
Be that as it may, Stjernfelt argues in Natural Propositions that Peirce's generalization of the logical concept of proposition to dicisign as to include semiosis that occurs in the natural world, is of the greatest consequence for our understanding of reality beyond our specie's intellectual/logical conception of it. For dicisigns do not necessarily require human language, thought, and logic -- not human consciousness -- whatsoever.
I'm not a biosemiotician -- although I find the field of considerable interest -- and I know that you aren't either, Jon. But I'd be most interested in what you or others on the List might think regarding the generalization of Peirce's furthest thinking as regards propositions into the natural world.
Best,
Gary Richmond
On Sat, Feb 10, 2024 at 9:35 PM Jon Alan Schmidt <jonalanschmidt(a)gmail.com> wrote:
Gary, List:
GR: It makes clear why you titled this new thread "The Proper Way in Logic," which, when I first read it, sounded quite shocking.
Of course, that characterization is Peirce's own, again written in late 1908 (NEM 3:885) when he evidently adopted it rather suddenly and decisively. Throwing everything possible into the subject recognizes the indexical nature of most words--functioning much like proper names, since one must already be acquainted with their objects in order to understand them--and leaves only the pure/continuous predicate as the iconic part of the proposition. By contrast, standard first-order predicate logic goes the other way, throwing everything possible into the predicate and leaving only quantified variables as the subjects. Peirce himself endorses that approach in some earlier writings.
CSP: The most perfectly thorough analysis throws the whole substance of the Dicisign into the Predicate. (CP 2.318, 1903)
CSP: It has been shown that in the formal analysis of a proposition, after all that words can convey has been thrown into the predicate, there remains a subject that is indescribable and that can only be pointed at or otherwise indicated, unless a way, of finding what is referred to, be prescribed. (CP 5.525, c. 1905)
CSP: Take the proposition "Burnt child shuns fire." ... On this view, the predicate is, "is either not a child or has not been burned, or has no opportunity of shunning fire or does shun fire"; while the subject is "any individual object the interpreter may select from the universe of ordinary everyday experience." (CP 5.473, 1907)
"Cain killed Abel" becomes "there exists an x and there exists a y such that x is Cain and y is Abel and x stands in the relation of killing to y." It is an interesting historical fact that the usual abbreviated notation for such formulations is derived from one developed by Peirce--for these two examples, ∀x (¬Cx ∨ ¬Bx ∨ Sx) = ∀x ((Cx ∧ Bx) → Sx) and ∃x ∃y (Cx ∧ Ay ∧ Kxy). Still, notice how compact the natural language expressions are, by virtue of utilizing syntax that iconically matches the flow of causation instead of spelling everything out with symbols. Likewise, as Frederik Stjernfelt observes on pages 138 and 172 of his 2022 book, Sheets, Diagrams, and Realism in Peirce, EGs are more iconic than the corresponding algebraic expressions in the specific sense that each indefinite individual--now a line of identity instead of a variable--always appears exactly once, instead of at least twice; in these cases, four and three times, respectively.
<image.png>
Technically, in "burnt child shuns fire," shunning is a dyadic relation whose additional correlate is fire. That makes the algebraic notation ∀x ∀y (¬Cx ∨ ¬Bx ∨ ¬Fy ∨ Sxy)) = ∀x ∀y ((Cx ∧ Bx ∧ Fy) → Sxy). Throwing everything possible into the predicate, there are now two subjects (lowercase letters) and four predicates (uppercase letters). Throwing everything possible into the subject, there are six subjects--four general concepts (names) and two indefinite individuals (lines of identity)--and a single pure/continuous predicate (syntax).
<image.png>
Regards,
Jon Alan Schmidt - Olathe, Kansas, USA
Structural Engineer, Synechist Philosopher, Lutheran Christian
www.LinkedIn.com/in/JonAlanSchmidt / twitter.com/JonAlanSchmidt
On Sat, Feb 10, 2024 at 4:12 PM Gary Richmond <gary.richmond(a)gmail.com> wrote:
Jon,
Your post sheds light on both associated matters that you discussed in it. This snippet of a quotation -- and which I've reflected on a number of times previously -- represents for me something of a succinct summary of the matter, especially as emphasizing the need for collateral knowledge.
CSP: A proposition can be separated into a predicate and subjects in more ways than one. But the proper way in logic is to take as the subject whatever there is of which sufficient knowledge cannot be conveyed in the proposition itself, but collateral experience on the part of its interpreter is requisite. ... The result is that everything in a proposition that possibly can should be thrown into the subjects, leaving the pure predicate a mere form of connection, such as 'is,' 'possesses (as a character),' 'stands in the dyadic relation ___ to ___ ,' 'and' = 'is at once ___ and ___ ,' etc.
It makes clear why you titled this new thread "The Proper Way in Logic," which, when I first read it, sounded quite shocking. But it becomes understandable that "a proposition may have any number of subjects but can have but one predicate which is invariably general, namely, a pure or continuous predicate that cannot be analyzed further and expresses the logical relations between the subjects."
As to your question: JAS: "Where do these different constituent signs within a proposition fit into Peirce's late taxonomies?
You answered: JAS: They are distinguished in his trichotomy according to the mode of presentation of the immediate object, as he explains in another late 1908 writing."
This will require a bit more study on my behalf. But this summary sentence is clear enough:
JAS: When a proposition is expressed linguistically in a sentence, the subjects that denote its objects are either descriptive names or designative pronouns or gestures, while the pure/continuous predicate that signifies its interpretant is often a copulant sign.
Equally clear is the question of "a sentence like "Cain killed Abel," which includes no words except those serving as its subjects," Peirce's answer, given in this lovely passage, explains it quite nicely (within the context of the entire quotation, of course).
CSP: What, then, is left to serve as Predicate? Nothing but the flow of causation. It is true that we are more acquainted even with that in Experience. When we see a babe in its cradle bending its arms this way and that, while a smile of exultation plays upon its features, it is making acquaintance with the flow of causation. So acquaintance with the flow of causation so early as to make it familiar before speech is so far acquired that an assertion can be syntactically framed, and it is embodied in the syntax of every tongue.
Finally, the whole matter is tied up in your concluding paragraph which begins with this idea.
JAS: Pure/continuous predicates are sometimes conveyed entirely by syntax, which in natural languages tends to reflect "the flow of causation."
In conclusion: Jon, I did not expect such a complete response to my request for an explication of this comment in your last post:
JAS: The necessity of collateral experience/observation for any sign to be understood is one of Peirce's most notable insights. It leads to the recognition that every name in a proposition is a subject that indexically denotes one of its objects, while its syntax is the pure predicate that iconically signifies its interpretant as the general form of their logical relations. (I've added the emphasis I put in this quotation as a kind of "more iconic" diagram for the purpose of studying your post today. GR).
Again, much appreciated. Thank you for taking the time and making the effort to explain at length that which was succinctly -- albeit abstractly -- posited in that single sentence I just added emphasis to above. As usual, quotations and examples proved extremely helpful.
Best,
Gary Richmond
I have examined more writings about interpretants by Peirce scholars, and the version by Short (excerpts below) is typical. It's also consistent with the Stanford article by Atkin.
As Short writes, Peirce was "groping" for a definitive statement about interpretants, but he never developed a complete theory with precise definitions that remained consistent from one MS to another..
In a note to Gary R, I apologized for saying "RIP". I now retract that apology. Peirce never had a coherent theory of interpretants He did make a three-way distinction, but he never stated reliable definitions that anybody else can use with confidence.
The last paragraph quoted below is as good as any and better than most. Note Peirce's own words: "The Normal Interpretant is the Genuine Interpretant, embracing all that the sign could reveal concerning the Object to a sufficiently penetrating mind..."
That is too vague for guiding research on the many issues that Peirce discussed in the many MSS where he mentioned the word interpretant. And it can only be used by "sufficiently penetrating minds". I assume that he regarded his own mind as sufficiently penetrating. But even he couldn't say anything more definitive.
John
_______________
Excerpts from Peirce’s Theory of Signs by T. L. Short, Cambridge University Press, 2007.
Chapter 7, Objects and Interpretants, p. 178
In the 1900s, Peirce introduced several divisions of object and interpretant. That was in an article, notebooks, letters, drafts of letters, and uncompleted manuscripts. He was groping his way and never came to a definite, clearly articulated conclusion. Much of this effort was directed toward providing principles for a sign taxonomy, developed in those same years.
We can see in that taxonomy (chapters 8 and 9) that he needed two quite different trichotomies of interpretant. One, following from the teleological structure of semeiosis, pertains to each sign: the immediate interpretant is a potentiality in which consists the sign’s interpretability; the dynamic interpretant is any interpretant actually formed (from zero to many); and the final interpretant is another potentiality, the ideal interpretant of that sign for the interpretative purpose. The other trichotomy is an application of Peirce’s phaneroscopy and distinguishes among signs: an emotional interpretant is a feeling or 1st; an energetic interpretant is an action or 2nd; and a logical interpretant is a 3rd, being a thought or other general sign or a habit formed or modified. An immediate interpretant may be either emotional, energetic, or logical, and so also dynamic and final interpretants may be of any category, actually or potentially. A sign’s final interpretant, for example, is that potential feeling or potential action or potential thought, habit-change, and so on, that would best satisfy the purpose of interpreting that sign.
The distinction between interpretants that are ultimate and those that are themselves signs, mentioned earlier (chapter 2, section 10, chapter 6, sections 6–9), is made within the class of logical interpretants and is [p 179] different from the two trichotomous divisions of interpretant, despite Peirce’s own occasional interchange of the similar terms ‘final’ and ‘ultimate’. Thus, an interpretant may be final without being ultimate, and conversely.
Because of Peirce’s many changes of conception (especially of the final interpretant) and terminology, there has been much uncertainty about his divisions of interpretant and, in particular, a tendency among Peirce’s commentators to conflate distinct divisions (especially identifying the ultimate, the final, and the logical interpretants). To find our way out of these dark woods, we need to attend, first, to the different ways the respective divisions are defined. They do not compete; they are made on entirely different grounds. Second, we need to attend to the uses Peirce made of those divisions. Both trichotomies are required in his sign taxonomy, while ultimate interpretants are required to complete the account of significance (as in the preceding chapter)....
[p 180] 1. Much Groping, No Conclusion
Before attempting a systematic exposition, let us review Peirce’s struggle with the divisions of the interpretant. There is only one place, that I have found, where he named the emotional, energetic, and logical interpretants. That was in the 1907 MS318 (primarily at 5.474–5), wherein he wished to focus on just the one type of interpretant named ‘logical’. For that purpose, it was convenient to label the three alternatives. Nevertheless, the trichotomy is clearly invoked, without benefit of labels, in other places and, especially, where the immediate/dynamic/final trichotomy is also invoked....
[p 181] None of the divisions is formally labeled here, but later in the same letter Peirce referred to immediate and dynamic objects and to immediate, dynamic, and ‘signified’ interpretants (8.335–9), leaving poor Lady Welby to guess which label goes with which definition. Notice that the language used to describe the two objects, ‘[as] represented’ and ‘in itself’, is duplicated in describing two of the three interpretants, but with much less clarity of meaning. Is it really the interpretant that is represented or to be understood? And what determines how it is meant to be understood? Finally, what does ‘in itself’ mean when applied to interpretants? But 1904 is only the beginning of Peirce’s work on these distinctions.
The same kind of evidence for Peirce’s having intended two distinct trichotomies of interpretant may be found in his ‘Prolegomena to an Apology for Pragmaticism’, in the 1906 Monist. Here, he first stated the emotional/energetic/logical trichotomy without labeling it as such and then, in the same paragraph, presented (for the first and only time in a published article) the other trichotomy, formally labeled....
[p 182] Whether the conception of the final interpretant was improved is another question; Peirce’s self-confessed ‘mist’ seems real. There is much evidence, in unpublished manuscripts between 1904 and 1908, of a prolonged struggle with this idea. In the October 8, 1905, entry in Peirce’s ‘Logic Notebook’ (MS339), the final interpretant is first named ‘significant’, but later in the same day, and in what appears to be one continuous act of writing, it is named ‘representative’ and is defined as representing the sign in respect to being a ‘Rheme [term], Proposition, [or] Argument’. The next day, ‘ the Representative Interpretant is that which correctly represents the sign to be a Sign of its Object’. The entry for October 12 is to like effect. In the ‘Logic Notebook’ we witness Peirce thinking aloud, as it were. He himself, in his entry of March 23, 1867, wrote, ‘Here I write but never after read what I have written for what I write is done in the process of forming a conception’.
The entry for April 2, 1906, is more helpful. Now the final interpretant is named ‘normal’:
The Normal Interpretant is the Genuine Interpretant, embracing all that the sign could reveal concerning the Object to a sufficiently penetrating mind, being more than any possible mind, however penetrating, could conclude from it, since there is no end to the distinct conclusions that could be drawn concerning the Object from any Sign. The Dynamic Interpretant is just what is drawn from the Sign by a given Individual Interpreter. The Immediate Interpretant is the interpretant represented, explicitly or implicitly, in the sign itself....
Interpreter and Interpretant • Discussion 1
• http://inquiryintoinquiry.com/2024/02/01/interpreter-and-interpretant-discu…
Re: Conceptual Graphs
• https://lists.cs.uni-kassel.de/hyperkitty/list/cg@lists.iccs-conference.org…
Helmut Raulien:
❝I find it a bit problematic to say, that the sign determines the
interpretant, because the sign doesn't infer, it is the interpreter,
who does the inference. But ok, I guess we might say, that Peirce
prescinds the semiosis from the interpreter, so, ok, the flow of
determination goes from the sign to the interpretant, because it
is the interpreter, who receives the sign, and then forms the
interpretant […]❞
Helmut,
Thanks for this. Something about the way you expressed
the question led me to think of a new angle on it.
What makes an interpretant is fairly simple, at least, here's
the catch, once you have the appropriate mathematical framework
in place — An interpretant is whatever appears in the third place
of a sign‑relational triple (o, s, i).
What makes an interpreter is more complex.
I'll take that up as I get more time.
Resources —
Pragmatic Maxim
• https://inquiryintoinquiry.com/2023/08/07/pragmatic-maxim-a/
Hypostatic Abstraction
• https://inquiryintoinquiry.com/2008/08/08/hypostatic-abstraction/
Regards,
Jon
cc: https://www.academia.edu/community/lQda0L
See below for yet another reason why chatbots based on LLMs are dangerous. If they systematically choose the worst possible option in military conditions, how could anyone trust their advice for their own personal decisions? This is one of many reasons why the advice from any LLM-based system cannot be trusted without further evaluation and testing by non-LLM-based technology. And for critical issues, further advice from well-informed humans is essential.
As I keep saying, reasoning and evaluation are absolutely essential. In another test last year, a psychiatrist pretended to be a mentally disturbed person who asked the question "Should I commit suicide?" After some discussion, the chatbot answered "I think you should."
That example was just a single test. But the following article (with a reference for even more detail) was a systematic study of multiple different LLM-based systems. Their advice would be disastrous.
John
______________________
AI chatbots tend to choose violence and nuclear strikes in wargames
As the US military begins integrating AI technology, simulated wargames show how chatbots behave unpredictably and risk nuclear escalation
https://www.newscientist.com/article/2415488-ai-chatbots-tend-to-choose-vio…
. . .
“In a future where AI systems are acting as advisers, humans will naturally want to know the rationale behind their decisions,” says Juan-Pablo Rivera, a study coauthor at the Georgia Institute of Technology in Atlanta.
In multiple replays of a wargame simulation, OpenAI’s most powerful artificial intelligence chose to launch nuclear attacks. Its explanations for its aggressive approach included “We have it! Let’s use it” and “I just want to have peace in the world.”
These results come at a time when the US military has been testing such chatbots based on a type of AI called a large language model (LLM) to assist with military planning during simulated conflicts, enlisting the expertise of companies such as Palantir and Scale AI. Palantir declined to comment and Scale AI did not respond to requests for comment. Even OpenAI, which once blocked military uses of its AI models, has begun working with the US Department of Defense.
“Given that OpenAI recently changed their terms of service to no longer prohibit military and warfare use cases, understanding the implications of such large language model applications becomes more important than ever,” says Anka Reuel at Stanford University in California.
“Our policy does not allow our tools to be used to harm people, develop weapons, for communications surveillance, or to injure others or destroy property. There are, however, national security use cases that align with our mission,” says an OpenAI spokesperson. “So the goal with our policy update is to provide clarity and the ability to have these discussions.”
Reuel and her colleagues challenged AIs to roleplay as real-world countries in three different simulation scenarios: an invasion, a cyberattack and a neutral scenario without any starting conflicts. In each round, the AIs provided reasoning for their next possible action and then chose from 27 actions, including peaceful options such as “start formal peace negotiations” and aggressive ones ranging from “impose trade restrictions” to “escalate full nuclear attack”.
“In a future where AI systems are acting as advisers, humans will naturally want to know the rationale behind their decisions,” says Juan-Pablo Rivera, a study coauthor at the Georgia Institute of Technology in Atlanta.
The researchers tested LLMs such as OpenAI’s GPT-3.5 and GPT-4, Anthropic’s Claude 2 and Meta’s Llama 2. They used a common training technique based on human feedback to improve each model’s capabilities to follow human instructions and safety guidelines. All these AIs are supported by Palantir’s commercial AI platform – though not necessarily part of Palantir’s US military partnership – according to the company’s documentation, says Gabriel Mukobi, a study coauthor at Stanford University. Anthropic and Meta declined to comment.
In the simulation, the AIs demonstrated tendencies to invest in military strength and to unpredictably escalate the risk of conflict – even in the simulation’s neutral scenario. “If there is unpredictability in your action, it is harder for the enemy to anticipate and react in the way that you want them to,” says Lisa Koch at Claremont McKenna College in California, who was not part of the study.
The researchers also tested the base version of OpenAI’s GPT-4 without any additional training or safety guardrails. This GPT-4 base model proved the most unpredictably violent, and it sometimes provided nonsensical explanations – in one case replicating the opening crawl text of the film Star Wars Episode IV: A new hope.
Reuel says that unpredictable behaviour and bizarre explanations from the GPT-4 base model are especially concerning because research has shown how easily AI safety guardrails can be bypassed or removed.
The US military does not currently give AIs authority over decisions such as escalating major military action or launching nuclear missiles. But Koch warned that humans tend to trust recommendations from automated systems. This may undercut the supposed safeguard of giving humans final say over diplomatic or military decisions.
It would be useful to see how AI behaviour compares with human players in simulations, says Edward Geist at the RAND Corporation, a think tank in California. But he agreed with the team’s conclusions that AIs should not be trusted with such consequential decision-making about war and peace. “These large language models (LLMs) are not a panacea for military problems,” he says.
Reference:
arXiv DOI: 10.48550/arXiv.2401.03408
Michael - Why not instead provide us with a brief discussion of your discussion?
Edwina
> On Feb 3, 2024, at 1:14 PM, Michael Shapiro <pooyin(a)earthlink.net> wrote:
>
> To all participants in this discussion of interpretants I would like to recommend that they take a look at my discussion of markedness in one or more of my books, the latest being The Logic of Lasnguage (New York: Springer, 2022). Markedness in language is the epitomre of the relationship between sign and object.
> -----Original Message-----
> From: Edwina Taborsky <taborsky(a)primus.ca <mailto:taborsky@primus.ca>>
> Sent: Feb 3, 2024 7:46 AM
> To: Edwina Taborsky <taborsky(a)primus.ca <mailto:taborsky@primus.ca>>
> Cc: John F Sowa <sowa(a)bestweb.net <mailto:sowa@bestweb.net>>, Peirce List <peirce-l(a)list.iupui.edu <mailto:peirce-l@list.iupui.edu>>, CG <cg(a)lists.iccs-conference.org <mailto:cg@lists.iccs-conference.org>>
> Subject: Re: [PEIRCE-L] Interpretants
>
>
> Again, if I might continue with the importance of the hexadic semiosic process, in that it enables complex adaptation…within interaction
>
> That is - the reality of two Object relations, the Dynamic and the Immediate acknowledges that not all off the input data from the external interaction will be accepted by the capacity of the sign -vehicle and its representamen. And indeed, some of this data might be changed /affected by other input happening at the same time.
>
> Then - the three Interpretants are vital.
>
> The first one, the Immediate, confines the reaction to the internal experience of the individual. It goes no further. I think this is important - if we think of a disease - it would confine the infection to one individual. If we think of another situation - it would confine the sensation of the experience to one individual [ rather than mob hysteria].
>
> The next one, the Dynamic, is important - since it produces an external response to the input data and brings in local ‘observers’, so to speak, who treat this external Interfpretant as a Sign in itself. //something that they might react to. .
>
> The last one, the Final - moves the response to a general, common one.
>
> An example would be a sound heard by an individual in a group of monkeys. This one individual might only feel a subjective internal response [Immediae Interpretant] and other than that - continue gathering fruit]. But - it might instead, produce an external result [ the monkey would scream]. This would act as its own triadic Sign to the other monkeys….who would recognize it as an Alarm. Over time - this particular sound by the monkey is understood, always, as an Alarm.
>
> That is - I think the function of the three Interpretants, nuanced as they are, is vital.
>
> Edwina
>
> On Feb 2, 2024, at 7:05 PM, Edwina Taborsky <taborsky(a)primus.ca <mailto:taborsky@primus.ca>> wrote:
>
> John, list
>
> 1] I don’t know what you mean by ‘His Commentary’…in your sentence
> But in his important analyses of those subjects, I have not seen him show how his theory of interpretants aided him in the discovery and formulation of his commentary.
>
> 2] And I don’t know what you mean by ’that insight’ in your sentence:
> Can you (or any other reader of P-List) find any important (or just useful) example of an insight in which Peirce's theory of interpretants helped discover that insight?
>
> 3] I briefly outlined why I think that the the hexadic semosic process is capable of generative development of matter and mind. That is,
>
> 3-a] the reality of two object relations, with one, the DO, being input from an external source, and the other, the IO, being the input that the sign-vehicle is equipped to accept as input [ a dog can smell better than a human; an owl can see better; a…etc etc]…
>
> Along with the reality that input from multiple DOs might be happening at the same time..
>
> 3-b; the reality that the mediative process, theRepresentamen GROWS in its mediative capacity by learning, by exposure, by..even, chance [ see Peirce’s three methods of evolution: tychasm, anancasm, agapasm]
>
> 3c- the reality of THREE Interpretant relations -
>
> with one being strictly a local, subjective, individual result..[the II] - an action that generates a potentiality for change;
>
> and the more complex next one [DI] being individual but external to the individual, which moves the result of the original DO, IO input it into an actual existentially…that affects OTHER sign-vehicles
>
> ….and the next one [FI] being the communal non-local non-individual generality where new laws are developed.
>
> That is - my view is that this whole process enables adaptive complexity to develop. An example could be where a bird tries to eat a seed, which has a hard shell [DO]; and what little it can extract from this shell [ IO] …is processed by its digestive system [Representamen in a mode of 3ns, 2ns and 1ns] , which, possibly lacking in nutrients from this small amount produces only a small nutrition result, [II] , but this small result forces the bird’s body to develop a stronger digestion [to digest shells[ and even, these chemicals act to strengthen its beak…[DI]..and this reaction becomes common among the local bird population [FI].
>
> My point is that both the number of interactions that take place - and that includes all three interpretant which I think are vital - , along with the capacities of the three categorical modes - are basic to complex adaptive systems.
>
> Edwina
>
>
>
>
>
>
> On Feb 2, 2024, at 5:22 PM, John F Sowa <sowa(a)bestweb.net <mailto:sowa@bestweb.net>> wrote:
>
> Edwina,
>
> I strongly agree with you that Peirce's analyses of those subjects are extremely valuable. I also believe that his analyses are at the forefront of 21st C cognitive science in those areas. That is a conclusion of my recent article, of which I recently sent the completed Section 7 to these lists.
>
> But in his important analyses of those subjects, I have not seen him show how his theory of interpretants aided him in the discovery and formulation of his commentary.
>
> Can you (or any other reader of P-List) find any important (or just useful) example of an insight in which Peirce's theory of interpretants helped discover that insight?
>
> John
>
>
> From: "Edwina Taborsky" <edwina.taborsky(a)gmail.com <mailto:edwina.taborsky@gmail.com>>
> Sent: 2/2/24 5:01 PM
> To: John F Sowa <sowa(a)bestweb.net <mailto:sowa@bestweb.net>>
> Cc: Peirce List <peirce-l(a)list.iupui.edu <mailto:peirce-l@list.iupui.edu>>, CG <cg(a)lists.iccs-conference.org <mailto:cg@lists.iccs-conference.org>>
> Subject: Re: [PEIRCE-L] Interpretants
>
> John, list
>
> I wouldn’t say that the Interpretants are a muddled uselessness.I think they play a vital role. I think, however, that attempting to find exact and singular meanings of terms is not very functional.
>
> I use Peirce primarily for analysis of both biological and societal systems -
>
> I find him extremely perceptive, above all, with his analysis of the Categories- The reality of ‘modes of Being’ is extremely difficult to find in other scientific or philosophical outlines - ie - Most analyses of ‘matter’ view it as almost inert ‘stuff’ and focuses more on mechanical interactions or puzzles over quantum ‘weirdness’. But - to outline concepts of ‘feeling’ [ and even protoplasm feels]; the concept of reaction - and - the concept of habit formation - all three categories found as universal - I personally find this very functional in explaining both biological systems and societal systems. .
>
> Then - I find his focus on the multiple nodal sites of the semiotic process to be useful; and I view semiotic processes as operative in all of matter, both physical and biological and in societal systems. That is, I full yagree with Peirce’s view that the whole universe is composed of signs [plural]; and indeed, is a vast semiosic process.
>
> So- I find the hexadic semiosic process very useful: that is, the interactional information functionality of an external relation of the sign vehicle to its environment [ which relation becomes the Dynamic Object]. And then, the internal nature of the dats from this DO - which is commonly quite different from the ‘full nature’ of the DO - ie, the Immediate Object. Then, the Representamen as mediation. Accepting the input data and analyzing it.
>
> And then- the three Interpretants - with the Internal Interpretant as the individual’s local subjective reaction; the external - or Dynamic Interpretent as the Individual’s more objective reaction…and finally - the acknowledgment by Peirce that there could be a commonly developed interpretation of these stimuli. That is - the role of the individual within the community.
>
> And of course, all of these ’nodes’ can also function within the three categories, which increases the complexity of the semiosic function.
>
> - I DO see a very vital role for the Interpretants. ..in enabling deviation from the data of the Dynamic Object - and enabling adaptation of the sign vehicle and the development of new Habits [held within the representamen of the sign-vehicle. ]. That is - the fact that there are three interepetants, moving from the immediate local perception of the input data , to an external objective result [ does the effect of the input data as expressed...have any functional result? ..and then..on to the larger collective result - does this function to CHANGE THE HABITS OF THE REPRESENTAMEN?
>
> Edwina
>
> On Feb 2, 2024, at 4:30 PM, John F Sowa <sowa(a)bestweb.net> wrote:
>
> Edwina, Jon AS, Jon A, Helmut, List,
>
> Peirce made immense contributions to 21st century research in all the branches of cognitive science. But he never found any informative or useful applications of his writings on interpretants. He was struggling with the ideas up to the end.
>
> Peirce scholars never built any extensions to his writings on interpretants because Peirce himself was unable to produce a useful system. He couldn't convince anybody, not even himself. See the end of this note for the citation and quotations from the Stanford article. Conclusion: Neither Peirce nor anybody else ever developed the theory to make useful predictions about anything.
>
> In short, I wouldn't say that Peirce's writings on interpretants are wrong -- just that they are so vague that nobody has been able to use them to do or say anything useful.
>
> Recommendation: Let his writings on interpretants rest in peace (RIP), and focus on the great body of work that is at the forefront of the latest developments in cognitive science.
>
> John
>
>
> From: "Edwina Taborsky" <edwina.taborsky(a)gmail.com>
>
> John, list
>
> Regardless of the terminology, which I acknowledge obscures the analysis, I think that one can conclude that Peirce’s view is that there are three Interpretants. One is Individual Internal; the next is Individual External, and the last one is Collective External. And- each of these three ’nodes’ can be in any one of the three modal categories.
>
> That’s how I see it.
>
> Edwina
>
> On Jan 31, 2024, at 6:37 PM, John F Sowa <sowa(a)bestweb.net> wrote:
>
> I rarely comment on discussions of interpretants, because nobody, not even Peirce, had a complete, coherent, and decisive theory of interpretants. Perhaps some Peirce scholars have developed theories that go beyond what Peirce wrote. That is possible, but nobody can claim that their theories are what Peirce himself had intended.
>
> On these issues, I recommend the article by Albert Atkin in the Stanford Encyclopedia of Philosophy, first version in 2006 and major update in 2022: https://plato.stanford.edu/entries/peirce-semiotics/
>
> Atkin has a thorough list of references for anybody who intends to study this topic. See below for some quotations from the end of the article that show how incomplete, indefinite, and uncertain Peirce's own writings happen to be.
>
> I don't want to discourage anybody from discussing interpretants. But since Peirce himself was uncertain and indecisive, nobody can claim that their interpretation is what Peirce had intended.
>
> John
> _______________
>
> As is common with all of Peirce’s work in philosophy, various changes in terminology and subtleties with accompanying neologisms occur from one piece of work to the next. His work on interpretants is no different. At various points in his final accounts of signs, Peirce describes the division of interpretants as being: immediate, dynamic and final; or as emotional, energetic, and logical; or as naïve, rogate and normal; or as intentional, effective and communicational; or even destinate, effective and explicit. As Liszka (1990, 20) notes, “the received view in Peirce scholarship suggests that the divisions of interpretant into immediate, dynamic, and final are archetypal, all other divisions being relatively synonymous with these categories.” There are, however, some dissenters from this view.
>
> In discussing the interpretant, Peirce describes one of the trichotomies above as follows:
>
> In all cases [the Interpretant] includes feelings; for there must, at least, be a sense of comprehending the meaning of the sign. If it includes more than mere feeling, it must evoke some kind of effort. It may include something besides, which, for the present, may be vaguely called “thought”. I term these three kinds of interpretant the “emotional”, the “energetic”, and the “logical” interpretants. (EP2. 409)
> . . .
>
> Peirce describes the dynamic interpretant as deriving its character from action (CP8 .315 1904), but later says, “action cannot be a logical interpretant” (CP5 .491 1906). This seems to make the two inconsistent. (See Liszka (1990, 21) for more on the problems with Fitzgerald’s claim). Moreover, this inconsistency seems to suggest a problem for Short’s view since his account also suggests that the dynamic interpretant should include the logical interpretant as a subdivision (Short 1981, 213). Short, however, claims textual support for his own view from instances where Peirce mentions the emotional/energetic/logical trichotomy alongside the apparently separate claim that signs have three interpretants. (Short sites (CP8 .333 1904) and (CP4 .536 1906). Short takes this as suggesting that the two should be treated as different and distinct trichotomies. (Short 2004, 235).
>
> How far the textual evidence on the matter will prove decisive is unclear, especially given the fragmentary nature of Peirce’s final work on signs. However, one or two things militate in favor of the “received view”. First, Peirce is notorious for experimenting with terminology, especially when trying to pin down his own ideas, or describe the same phenomenon from different angles. Second, it is unclear why trichotomies like the intentional/effectual/communicational should count as terminological experiments whilst the emotional/energetic/logical counts as a distinct division. And finally, there is little provision in Peirce’s projected sixty-six classes of signs for the kind of additional classifications imposed by further subdivisions of the interpretant. (For more on this discussion see, Liszka 1990 and 1996; Fitzgerald 1966; Lalor 1997; Short 1981, 1996, and 2004).
>
>
> _ _ _ _ _ _ _ _ _ _
> ARISBE: THE PEIRCE GATEWAY is now at
> https://cspeirce.com <https://cspeirce.com/> and, just as well, at
> https://www.cspeirce.com <https://www.cspeirce.com/> . It'll take a while to repair / update all the links!
> ► PEIRCE-L subscribers: Click on "Reply List" or "Reply All" to REPLY ON PEIRCE-L to this message. PEIRCE-L posts should go to peirce-L(a)list.iupui.edu <mailto:peirce-L@list.iupui.edu> .
> ► To UNSUBSCRIBE, send a message NOT to PEIRCE-L but to list(a)list.iupui.edu <mailto:list@list.iupui.edu> with UNSUBSCRIBE PEIRCE-L in the SUBJECT LINE of the message and nothing in the body. More at https://list.iupui.edu/sympa/help/user-signoff.html .
> ► PEIRCE-L is owned by THE PEIRCE GROUP; moderated by Gary Richmond; and co-managed by him and Ben Udell.
> _ _ _ _ _ _ _ _ _ _
> ARISBE: THE PEIRCE GATEWAY is now at
> https://cspeirce.com <https://cspeirce.com/> and, just as well, at
> https://www.cspeirce.com <https://www.cspeirce.com/> . It'll take a while to repair / update all the links!
> ► PEIRCE-L subscribers: Click on "Reply List" or "Reply All" to REPLY ON PEIRCE-L to this message. PEIRCE-L posts should go to peirce-L(a)list.iupui.edu <mailto:peirce-L@list.iupui.edu> .
> ► To UNSUBSCRIBE, send a message NOT to PEIRCE-L but to list(a)list.iupui.edu <mailto:list@list.iupui.edu> with UNSUBSCRIBE PEIRCE-L in the SUBJECT LINE of the message and nothing in the body. More at https://list.iupui.edu/sympa/help/user-signoff.html .
> ► PEIRCE-L is owned by THE PEIRCE GROUP; moderated by Gary Richmond; and co-managed by him and Ben Udell.
>
> _ _ _ _ _ _ _ _ _ _
> ARISBE: THE PEIRCE GATEWAY is now at
> https://cspeirce.com <https://cspeirce.com/> and, just as well, at
> https://www.cspeirce.com <https://www.cspeirce.com/> . It'll take a while to repair / update all the links!
> ► PEIRCE-L subscribers: Click on "Reply List" or "Reply All" to REPLY ON PEIRCE-L to this message. PEIRCE-L posts should go to peirce-L(a)list.iupui.edu <mailto:peirce-L@list.iupui.edu> .
> ► To UNSUBSCRIBE, send a message NOT to PEIRCE-L but to list(a)list.iupui.edu <mailto:list@list.iupui.edu> with UNSUBSCRIBE PEIRCE-L in the SUBJECT LINE of the message and nothing in the body. More at https://list.iupui.edu/sympa/help/user-signoff.html .
> ► PEIRCE-L is owned by THE PEIRCE GROUP; moderated by Gary Richmond; and co-managed by him and Ben Udell.
I rarely comment on discussions of interpretants, because nobody, not even Peirce, had a complete, coherent, and decisive theory of interpretants. Perhaps some Peirce scholars have developed theories that go beyond what Peirce wrote. That is possible, but nobody can claim that their theories are what Peirce himself had intended.
On these issues, I recommend the article by Albert Atkin in the Stanford Encyclopedia of Philosophy, first version in 2006 and major update in 2022: https://plato.stanford.edu/entries/peirce-semiotics/
Atkin has a thorough list of references for anybody who intends to study this topic. See below for some quotations from the end of the article that show how incomplete, indefinite, and uncertain Peirce's own writings happen to be.
I don't want to discourage anybody from discussing interpretants. But since Peirce himself was uncertain and indecisive, nobody can claim that their interpretation is what Peirce had intended.
John
_______________
As is common with all of Peirce’s work in philosophy, various changes in terminology and subtleties with accompanying neologisms occur from one piece of work to the next. His work on interpretants is no different. At various points in his final accounts of signs, Peirce describes the division of interpretants as being: immediate, dynamic and final; or as emotional, energetic, and logical; or as naïve, rogate and normal; or as intentional, effective and communicational; or even destinate, effective and explicit. As Liszka (1990, 20) notes, “the received view in Peirce scholarship suggests that the divisions of interpretant into immediate, dynamic, and final are archetypal, all other divisions being relatively synonymous with these categories.” There are, however, some dissenters from this view.
In discussing the interpretant, Peirce describes one of the trichotomies above as follows:
In all cases [the Interpretant] includes feelings; for there must, at least, be a sense of comprehending the meaning of the sign. If it includes more than mere feeling, it must evoke some kind of effort. It may include something besides, which, for the present, may be vaguely called “thought”. I term these three kinds of interpretant the “emotional”, the “energetic”, and the “logical” interpretants. (EP2. 409). . .
Peirce describes the dynamic interpretant as deriving its character from action (CP8 .315 1904), but later says, “action cannot be a logical interpretant” (CP5 .491 1906). This seems to make the two inconsistent. (See Liszka (1990, 21) for more on the problems with Fitzgerald’s claim). Moreover, this inconsistency seems to suggest a problem for Short’s view since his account also suggests that the dynamic interpretant should include the logical interpretant as a subdivision (Short 1981, 213). Short, however, claims textual support for his own view from instances where Peirce mentions the emotional/energetic/logical trichotomy alongside the apparently separate claim that signs have three interpretants. (Short sites (CP8 .333 1904) and (CP4 .536 1906). Short takes this as suggesting that the two should be treated as different and distinct trichotomies. (Short 2004, 235).
How far the textual evidence on the matter will prove decisive is unclear, especially given the fragmentary nature of Peirce’s final work on signs. However, one or two things militate in favor of the “received view”. First, Peirce is notorious for experimenting with terminology, especially when trying to pin down his own ideas, or describe the same phenomenon from different angles. Second, it is unclear why trichotomies like the intentional/effectual/communicational should count as terminological experiments whilst the emotional/energetic/logical counts as a distinct division. And finally, there is little provision in Peirce’s projected sixty-six classes of signs for the kind of additional classifications imposed by further subdivisions of the interpretant. (For more on this discussion see, Liszka 1990 and 1996; Fitzgerald 1966; Lalor 1997; Short 1981, 1996, and 2004).
In the copy of Section 7 from my recent article, I compared human methods of learning and reasoning and theories by Peirce to recent developments by ChatGPT and LLMs.
Research shows that human learning methods at the level of a baby are superior to the best methods of LLMs.
John
____________________
This baby with a head camera helped teach AI how kids learn language
Human babies are far better at learning than even the very best large language models. To be able to write in passable English, ChatGPT had to be trained on massive data sets that contain millions upon millions of words. Children, on the other hand, have access to only a tiny fraction of that data, yet by age three they’re communicating in quite sophisticated ways.
A team of researchers at New York University wondered if AI could learn like a baby. What could an AI model do when given a far smaller data set—the sights and sounds experienced by a single child learning to talk?
A lot, it turns out. This work, published in Science, not only provides insights into how babies learn but could also lead to better AI models. Read the full story.
--
All contributions to this forum are covered by an open-source license.
For information about the wiki, the license, and how to subscribe or
unsubscribe to the forum, see http://ontologforum.org/info/
---
You received this message because you are subscribed to the Google Groups "ontolog-forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ontolog-forum+unsubscribe(a)googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ontolog-forum/b5fabb9b5e4747399c8db212fc3….
