The Influence of Syntax on Updating Belief Centrally and Peripherally

A Comparison of Belief Updating in English, German and Russian

 

Henry Whittlesey Schroeder

 

Abstract on relevance to perypatetik project

This paper argues that the cognitive architecture of all peoples is identical, but language impacts the degree to which specific parts of the architecture are used. Syntax and conventions in English facilitate belief updating in central cognitive networks, while in German and Russian they encourage more peripheral network updating. The greater updating of beliefs via the central cognitive network in English explains the values and norms that predominate in the leitculture of pragmatism in America. Central belief updating is an all-or-nothing approach to beliefs that mirrors pragmatists’ unipolarity (materialism, consumption), focus on ends and success (rather than processes), proselytizing central beliefs, commitment to obligations and consistency, and limited compassion – values and norms piggybacking on the absence of access to peripheral networks. For romantics, it is largely the reverse: they are constantly negotiating between core and peripheral beliefs, which drives them to reject unipolar materialism and consumption, embrace the process of living or producing, indulge in the diversity of views (without advocacy), choose freedom (to switch between networks) over the obligation of consistency, and accept their polarized existence (see Peripatetic Alterity, chapters 2, 3 and 6).

Abstract on paper

With a model that divides cognition into central and peripheral networks, one might ask whether and under what conditions a (new) belief endorsed will update centrally rather than peripherally or vice-versa. This paper aims to show that one condition is the syntax of the language in which the (new) belief is processed.

I will argue that syntax and convention in (American) English (e.g., limited scrambling, right-branching structure (SVO), theme/rheme placement constraints) facilitate ceteris paribus either the rejection of a new belief or its endorsement in the central cognitive network (without mediation via peripheral networks). By contrast, syntax and convention in German and Russian (e.g., scrambling (both), left-branching (German) or dual-branching (Russian), theme/rheme flexibility (both)) establish ceteris paribus a pattern of updating where new beliefs are rejected less and frequently allocated to peripheral rather than central cognitive networks. The upshot of these differences is the polarization of (expressed) American cognition due to reduced updating via peripheral networks relative to German and Russian cognition with satellites of alternative, competing beliefs circling peripherally around central ones.

 

Prolegomenon

An (American) English-language bias in cognitive studies is well documented (Amici et al. 2019; Arnett 2008; Benitez-Burraco 2025; Blasi et al. 2022; Thalmayer et al. 2021). Many of these studies are either meta-analyses on the current state of research with few references to specific scientifically tested hypotheses or fail to lay out the practical or theoretical implications of less biased approaches beyond platitudes such as “cultures differ.”[1] In this paper, I head down the latter road, taking the theoretical fork on it.

My position is that specifiable, predictable differences in mental processing can be accounted for in part by divergences in conventional syntax from one language to another. A couple of examples to get us started: the common appearance of the verb in the finite position, as in German frequently and Russian when desired, is rare in English; while the possibility of placing the subject determiner phrase (subject-DP) in the finite position, as often done in the opening sentence of Russian newspaper articles, is even rarer in English. If a mental processing parser is habituated to hearing prepositional phrases or comparative functional heads while waiting for a subject- or object-DP or verb phrase (VP), it will differ from a parser knowing only a convention with all the subject- and object-DPs as well as the VP near to the sentence-initial position and uninterrupted by prepositional phrases or comparative functional heads. Simply put, there will be some difference between minds that regularly produce and process these clauses[2]:

(i) English:

Because the woman reads the journal rather than the newspaper on Wednesday.

(ii) German:

Because the woman on Wednesday rather than the newspaper the journal [obj.-DP] reads.

[weil die Frau am Mittwoch anstatt der Zeitung die Zeitschrift ließt.]

(iii) Russian:

Because the woman on Wednesday rather than the newspaper reads the journal [obj.-DP] reads. [потому что женщина в среду наместо газеты читает журнал.]

So far, so determinant are the consensus effects of syntax, I hope.

In this paper I attempt to go a bit further, however: namely, to understand the impact of syntax on belief updating to central and/or peripheral cognitive networks, and the consequences for fragmented cognition.

To ground the findings, I rely on two models: the initial automatic acceptance of beliefs followed by endorsement/rejection according to Mandelbaum’s interpretation of Spinoza (Mandelbaum 2014, 2016) along with my addition of the empirical-rational model of Kant (section one); and, in section two, the fragmented updating of beliefs (Petty and Wegener 1999; Bendaña and Mandelbaum 2021; Mandelbaum 2019, Sharot et al. 2023).

With this theoretical foundation in place, in section three, we look at discounting cues to see their effect on the endorsement of automatically accepted beliefs and the assignment of a (new) belief to a central or peripheral network. Here I draw an analogy between the function of antecedent and subsequent discounting cues for propositions on the one hand and the negation operator antecedent (in Russian) and subsequent (in German and English) to the copula and lexical or auxiliary verb in grammar on the other.

In section four, I contrast right-branching English with its relatively fixed word order and limitations on theme/rheme placement to the dual or left-branching structure of Russian and German, both with extensive scrambling and theme/rheme placement variability.

The paper culminates in section five with the consequences: namely, the syntax of English influences the parser to either reject the (new) belief or assign the updating to a central network; whereas the syntax of German and Russian facilitate the updating of peripheral networks, with fewer (new) beliefs rejected or allocated to a central network.

In conclusion, I discuss further evidence of these cultural differences and two objections, which in turn tie these findings to our contemporary social dynamics.

Ultimately, if my argument succeeds, it confirms the merits of both holistic theories of the mind, such as Quine’s, and fragmented ones such as Petty and Cacioppo’s or Mandelbaum’s, albeit relative to the processing language.

1. Road sign: the Spinozan theory of belief within a broadly Kantian framework

In The Ethics, Benedict de Spinoza presents a tripartite conception of knowledge or the mind inherited from the Middle Ages: reason (ratio), intuition and imagination. The intellect, which “perceives things through their first causes, and which is the same in all men” explains, for Spinoza, “why the mind, from the thought of one thing, immediately passes to the thought of another, which has no likeness to the first: as, for example, from the thought of the word pomum a Roman will immediately pass to the thought of the fruit [viz. an apple]” (Spinoza: EII/107). Reason gives humankind “common notions and adequate ideas of the properties of things”; intuition proceeds from an adequate idea of nature to adequate knowledge of the essence of things; and the imagination uses signs and words to form ideas of them (EII/122). In postulate 41, Spinoza states that only knowledge of the imagination can be false, whereas knowledge of reason and intuition is necessarily true (EII/122). Critically, “it is the nature of reason to perceive things truly”, that is, not as contingent but as necessary (EII/125).

The Spinozan model of belief formation is often contrasted with the Cartesian model (Mandelbaum 2014, Pion et al., Schiller et al.). In the Cartesian conception, the acceptance or rejection of a proposition is withheld, whereas in the Spinozan theory a proposition is accepted or taken up automatically upon its consideration, before it is either endorsed or rejected.[3] In the Spinozan model, as Pion et al. note, “acceptance, on the one hand, and rejection and endorsement, on the other must be accomplished via different mechanisms, one [acceptance] that is automatic and effortless, and one [endorsement and rejection] that is active and effortful” (Pion et al.: 3).

The Spinozan model for belief updating is faced with the risk that non-propositions, i.e., propositions without an empirical grounding, could be endorsed as truth-apt. For example, a belief in a supernatural God could be accepted and endorsed because knowledge, i.e. a propositional belief, of reason is always true, whereas the imagination, i.e. a world-sensitive belief, may be false (Spinoza: EII/122). If a broadly Kantian framework is supplemented, the endorsement of the supernatural is not possible. In Kant, belief acquisition also materializes automatically when the faculty of sensibility gives a representation to the intuition[4]. For example, the moment I take up an object such as a house, I have a belief. In Kant, however, this belief must be grounded in both the empirical world (verifiable in the language of the Logical Positivists) and in the rational structure (reproduced and synthesized in the faculty of sensibility and the faculty of the imagination before the acquisition of a category).[5]

The empirical basis, synthesis and thus possibility of applying a category constitute a “cognition object” for Kant. There are many states of affairs and words communicated that do not manifest themselves as representations given to the intuition and faculty of the imagination either because we are entirely oblivious to them or because a synthesis does not take place.[6] One can see this phenomenon in particular as one gets older: behaviors, expressions, relations previously unknown become part of active synthesized representations. But for whatever belief is endorsed, the representation must both be possible to experience and be assigned a category so as not to (erroneously) think (about) representations of entities that do not exist (e.g., God).[7]

Just as a belief must be propositional for it to be reason-responsive and possibly update a cognitive network, it must also be a cognition object for both the speaker introducing the (new) belief and the agent receiving it. Only with such truth-apt objects will the agent respond by using reason that will determine whether the (new) belief updates centrally or peripherally. With the synthesis of the representation of the cognition object, an automatically accepted belief is endorsed by the application of a category. Essentially, the assignment of a category constitutes acceptance in the Spinozan sense; lack thereof would be non-acceptance in the Spinozan sense and, equivalently, non-synthesis in the Kantian sense (something we are unaware of with both). Following acceptance/synthesis, a Kantian category is applied, and the (new) cognized belief object is either rejected or endorsed centrally or peripherally.

2. The map: central and peripheral cognitive networks in general

Now that we have reviewed how accepted propositional beliefs are endorsed in the Spinozan-Kantian model, we must address central and peripheral cognitive networks to understand the differences in central or peripheral belief updating due to the syntax of a language (sections three to five).

On one influential view, that of Quine, the endorsement of (new) individual beliefs (facts) causes a holistic shift in central cognition (theory/conceptual scheme). He argued that the endorsement of new facts and thereby their uptake into our conceptual scheme requires a holistic adjustment because the theory [i.e., belief] component cannot be reduced to the factual component(s): “A conflict with experience at the periphery occasions readjustments in the interior of the field” (Quine 1951: 38-9). The factual components, the periphery, in rival systems may even be contrary in truth value (Quine 1960: 73). The facts are not implied by any one of the component statements of the theory rather than another because the component statements do not have empirical meanings, but a sufficiently inclusive portion of theory does (Quine 1969: 79, 81-2).

An alternative approach divides cognition into a central and peripheral network or system. Beliefs can update either centrally or peripherally. One such model coming from the field of social psychology, the Elaboration Likelihood Model (ELM), was developed in the 1980s by Petty and Cacioppo. Attitudes (Beliefs) are changed either via the central or peripheral route (Petty et al. 1981: 847-8). A change via the central route involves relatively extensive and effortful information-processing activity aimed at scrutinizing and uncovering the central merits of the issue or advocacy (Petty et al. 1999: 43). In the case of updating via the peripheral route, less cognitive effort is required and information scrutiny is reduced (Petty et al. 1981: 848; Petty et al. 1999: 43). Although attitude change can occur via the peripheral route through classical conditioning, self-perception and heuristics, such belief updating is postulated to be weaker than attitude change via the central route (Petty et al. 1999: 43).[8] On these grounds, the authors suggest the possibility of two or more beliefs (“evaluative predispositions” or “attitudes” in their terms) toward a given attitude (belief) object (ibid 63).

In philosophy, dual networks of redundant (competing) beliefs have also been conceived as a psychological immune system defending itself despite another belief system recognizing a contradictory position (Mandelbaum 2019). Based on Batson’s experiment (1975) where religious believers who held the antecedent belief that Christ was the son of God, but also accepted a subsequent discounting cue, namely an article that proved Christ was not the son of God, Mandelbaum identifies a process where the antecedent religious believed the counterattitudinal article to be true and increased their belief in the face of evidence that they took to be disconfirming (150-1). In such a case, the belief that p must be on one network and the belief that ~p on another.[9]

Belief redundancy can also be broken down by voicing/action and thought – which becomes relevant to fragmented cognition, particularly in an English-language context, to ensure dispositional architectural identity of cognition across languages – because voicing or acting on a belief seems to result in a central update, while only thinking about one may allow for a peripheral update. In a slightly, but importantly different study, Festinger and Carlsmith found that if you induce someone to do or say something that is contrary to their private opinion, there will be a tendency for them to change their opinion to bring it in line with what they have done or said (Festinger and Carlsmith 209-10). Festinger and Carlsmith’s subjects acted, while Mandelbaum’s merely thought. Does speech or action mean that a subject has allocated more cognitive resources to the new belief, so attitude change is taking place on the central route, whereas disconfirming evidence regarded as true is assigned to a peripheral route that competes with the central route or psychological immune system? In other words, do Festinger and Carlsmith’s subjects induced to explain that a boring task is fun for $1 actually update their belief and rate the task as fun to a significantly greater extent than the control group and subjects in the $20 condition (ibid 207) precisely because of the relatively extensive and effortful information-processing activity aimed at scrutinizing and uncovering the central merits of the issue or advocacy (Petty and Wegener 43)? By contrast, Mandelbaum’s believers merely read an article, acknowledge its truth, but do not engage in any speech or action in defense of the evidence disconfirming their belief. Lefebvre et al. (2024) arrived at a similar conclusion in experiments on opinion polarization: independently of echo chambers and/or filter bubbles, the repetitive expression of one’s attitude is linked to the increase of the attitude’s extreme nature: “each expression opportunity activates and orients one’s confirmation bias in a way that amplifies the polarization of their attitudes, whereas listening without expression may not” (Lefebvre et al. 2024: 7).

What we will see in section five, inter alia, is that acting on or expressing a belief is language-independent, either attesting to or causing a central network belief update. This conclusion must be drawn from Festinger and Carlsmith as well as Lefebvre and Deroy. Accordingly, Quine’s holism applies to the endorsement of a structured belief that is acted on is language-independent. However, the endorsement of a (new) belief may only be in thought, i.e., a cognition object. Under these intellectual circumstances, i.e., when a belief is only thought, beliefs may update centrally or peripherally, and, as I will propose, the route of updating in such cases will be influenced by the syntax in a given language. Furthermore, analogous empirical work showing how this might work exists from experiments on the effect of discounting cues and the sleeper effect.

3. Signposts: belief and the negation operator

Not only does the analogy of a discounting cue to a negation operator serve as a model for how differences in syntax can influence the updating of divergent cognitive networks, but it also shows how belief updating involves only parts of sentences rather than whole ones.

3.1 Discounting cues antecedent and subsequent to a message function like the negation operator to a verb (English, German and Russian)

The sleeper effect has documented how a discounting cue is forgotten over time (Hovland et al. 1951; Pratkanis et al. 1988; Foos et al. 2016; Petrocelli et al. 2022). In one form, with negation operators, a negated predicate such as stingy will come to be associated with the subject, causing the positive valence of “he is not stingy” to turn negative when the “discounting cue” is forgotten. The agent hearing the sentence eventually recalls “he is stingy.”

In English syntax, the negation operator or discounting cue generally follows the copular verb “be” or the auxiliary required for lexical verbs such as “work,” “read,” “write,” etc.[10] For example, we say, in English, “he does not work” and “she is not stingy” rather than “he not work(s),” respectively “she not is stingy” as one often does in a language such as Russian.

To determine a possible difference between an antecedent and subsequent discounting cue, Pratkanis et al. (1988) manipulated their position by testing whether the provision of discounting information before or after the message induced a greater sleeper effect. The findings demonstrated that the sleeper effect occurred with greater frequency in the case of discounting information subsequent to the message (Pratkanis et al.: 213-14). Furthermore, the authors also noted that a discounting cue provided before the message did not reliably produce a sleeper effect (Pratkanis et al: 214).

Petrocelli et al. (2022) confirm a sleeper effect with a discounting cue subsequent to the message in their experiments on attitude change.[11] Students exposed to a gluten-free pizza promotion and learning subsequently that the content of the promotion was a lie or bullshit were more likely to exhibit the sleeper effect after a 14-day (delayed) assessment than the control group or students receiving the discounting cue antecedently: “attitudes generally became more positive about the pizza from the immediate assessment… to the delayed assessment” (Petrocelli et al. 2022: 415).

3.2 Position of negation operator as analogous to post- and pre-message discounting cues

The negation operator in English sentences adopts a subsequent position vis-à-vis the copular verb “be” and auxiliary verbs (“do” and “have”) required for negation. As such, it is analogous to a subsequent discounting cue vis-à-vis the message in the aforementioned experiments. The grammatically standard position of the negation operator[12] varies from language to language, however. In English, German and Russian, we have this inexhaustive list of options:

	English	German	Russian
Copular verb + predicate	She is not stingy She was not stingy	Sie ist nicht geizig Sie war nicht geizig (same as in English)	Она не скупа (= She not [is] stingy)[13] Она не была скупой (= She not was stingy)
Lexical verb	She does not work She did not work	Sie arbeitet nicht Sie hat […][14] nicht gearbeitet (or equivalently: sie arbeitete nicht)	Она не работает (= She not work) Она не работала (= She not worked)

First, a few comments are in store: (i) the position of the negation operator, as seen with the copular verb “be,” is either after the verb (as in English and German) or before the verb (as in Russian[15]); (ii) with lexical verbs, the negation operator in Russian precedes the verb in both the present and past tense; (iii) auxiliary verbs complicate the analysis in English and in the past tense of German (Perfekt): the auxiliary verbs “do” or “have” are generally necessary for English negation (their counterparts “haben” and “sein” (and “werden”) are required for the past tense (Perfekt) in (primarily) spoken German), with the auxiliary preceding the negation operator in English and German, but with the lexical verb following shortly thereafter in English. In German the negation following the auxiliary may appear much later in the sentence and often immediately before the lexical verb in the finite position. For example, “sie hat vor vier Tagen die Arbeit nicht gemacht” [“she [“has” – aux. verb] four days ago the work [obj.-DP] not done”].[16]

Second, to simplify our analysis here, I will solely speak of negation operators with copula verbs to avoid the complications of auxiliary verbs, restricting the scope to the pre- and post-verb negation operators. If evidence of the sleeper effect found by Pratkanis et al., Petrocelli et al., and Foos et al. is more pronounced with discounting cues following the message, then messages in English and German should be more liable to communication decay due to a (communication-receiving) agent’s initial processing of “be” followed by the negation operator “not.” Over time, as the experiments with discounting cues above showed, the subsequent negation operator will be forgotten. By contrast, those same experiments (especially Pratkanis et al., and Foos et al.) found that a cue before the message resulted in (only) a moderate opinion change and (only) a slight dissipation of that opinion after a period of six weeks (Pratkanis)[17] or no change (Foos). Accordingly, due to the verb-preceding position of the negation operator, the sleeper effect should be less prominent in Russian than in German or English.

One possibility accounting for a less prominent sleeper effect with antecedent discounting cues on the Spinozan-Kantian model may be that beliefs update on a peripheral rather than central cognitive network. Discounting cues preceding the message act like a warning about the new belief: if it is not rejected outright, its endorsement will be allocated to a peripheral network subject to future confirmation or reconsideration, but not to a central system. Analogously, but even more expansively due to the constant exposure to syntax, the negation operator preceding the verb (as in Russian) encodes a practice of initial peripheral network updating: communicators and recipients of communication repeatedly encounter negation operators like “she not [is] p” and, if endorsed, may not change their central belief about “her”, but will keep it in mind peripherally. Furthermore, this practice may even be expanded to propositions without negation operators – any endorsement initially unfolds peripherally.[18]

The updating of the central cognitive network in English and German, according to the above line of reasoning, is trained ceteris paribus for an endorsed belief because a subsequent discounting cue, even if not present, always lurks in the background due to syntax. Any message is liable to the sleeper effect. An endorsed (new) belief updated to a peripheral network might be forgotten like a discounting cue, so a new belief must enter the central cognitive network where it is reinforced through elaboration, strength, persistence, durability and other features of strongly held beliefs (which override any sleeper effect). The practice of losing a piece of information like the negation operator inculcated by a syntax with discounting cues subsequent to the message sidelines peripheral networks. There are only two options: rejection or central updating: if a belief aimed at central updating is in conflict with others in the central cognitive network, it will presumably be rejected; should it not be in conflict, as in, say, the example with gluten-free pizza, it will be incorporated into the central system. In either case, the build-out of an elaborate dual central and peripheral cognitive network is undermined by the position of the negation operator. In English, this dynamic is compounded by the next three features of syntax we will examine, but, showing how things are not all equal (ceteris paribus), these other constituents will shift German belief updating closer to Russian.

4. Signposts in English, German and Russian: linguistic relativism

So far we have seen how the Spinozan-Kantian model provides a model for belief endorsement/rejection grounded on the manifestation of beliefs in the world as entities capable of a reason-based, i.e., rational, response due to the production of representations/appearances, their synthesis in the imagination and assignment of a category by the faculty of understanding (section one); the possibility of updating beliefs on a central or peripheral cognitive network (section 2); and divergence in cognition as a result of syntactic structure, namely, the position of the discounting cue/negation operator (section three).

Before I explain the influence of syntax on belief updating, specifically the divergence between English syntax on the one hand and Russian and German on the other, we need to understand a few specific features of English, German and Russian that will ground the greater peripheral network updating of German and Russian speakers relative to their English counterparts.

The idea that language may impact thought processes has been around since at least Humboldt’s theory of divergence in languages as expressions of different worldviews (Trabant 25). This position now referred as “linguistic relativity” or the “Sapir-Whorf” hypothesis was developed by the linguist Edward Sapir and his student Benjamin Lee Whorf (Pütz/Verspoor ix). In recent years, cognitive scientists have attempted to study the influence of syntactic structures of language on the mind. One experiment involving multiple left-branching and right-branching languages found that left-branch speakers were better than right-branch speakers at recalling initial stimuli, i.e., the ability to retain initial information in a sentence, and an overall impressive performance by left-branching language speakers in highly educated countries (Japan and Korea) (Amici et al. 2024: 1, 6-8).

This picture of mental representation is grounded on the following syntactic and convention-based differences between American English on the one hand and German and Russian on the other[19]:

	English	German	Russian
Scrambling language (word order freedom)	No[20]	Yes[21]	Yes[22]
Left-branching or right-branching[23] or both	Right-branching[24]	Both, but considered left-branching[25]	Both[26]
Type of language	SVO	SOV	SVO[27], VSO[28]
Theme/rheme flexibility	Limited	Extensive[29]	Unlimited

What we will encounter below is that American English differs from German and Russian with regard to scrambling, branching/type of language (with attendant type of language), and theme/rheme flexibility.[30]

4.1 Syntax of American English

4.1.1 Scrambling

English is considered a right-branching subject-predicate or subject-verb-object language (SVO) with low tolerance for word order freedom (“scrambling”).[31] The slots in which syntactic elements can be placed is so limited that Fichtner charts out all the possibilities prior to the lexical verb (subject, intraverbal adverb (flexible) auxiliary) and following it (nominal, verb particle, verb complement, terminal adverb), something he does not (and cannot) do for German (Fichtner: 65).

4.1.2 Branching / Type: SVO

In the SVO language of English, the verb and object form one unit, with adverbials[32] attaching to the left or right of this unit. (Müller 24-5). For example, English does not allow orders in which the object appears immediately before the finite verb (Müller 19), i.e., one cannot say: “I this book give to you now.” The fundamental pattern is “noun-verb-noun” in English (Fichtner 99), but it can be any combination (noun-verb-noun, noun-noun-verb, verb-noun-noun) in German.

4.1.3 Theme/Rheme

The aforementioned quasi-rules for syntax, coupled with the lack of morphological change in nouns and their articles, constrain the options for phrases acting as theme or rheme. In English, it is difficult for the subject determiner phrase (subject-DP) to assume the function of rheme due to the subject-DP’s regular placement at or near the sentence-initial position while the rheme tends to appear closer to the finite position (especially in the event of anaphoric reference in the subsequent sentence). This syntactic requirement encourages object-determiner phrases (object-DPs) to assume the rheme function by virtue of being the sole alternative, with subsequent prepositional phrases or comparative functional heads appearing after the object-DP because of Fichtner’s aforementioned fundamental pattern of noun-verb-noun (SVO).

4.2 Syntax in German and Russian

4.2.1 Scrambling

German and Russian are considered “scrambling languages” in that there exists a process for transforming basic word orders into alternative ones (Bailyn 293).[33] In Russian, all six basic SVO word orders are grammatical (ibid 292; 294). Prepositional phrases, subject- or object-DPs or even comparative functional heads in German or Russian may function as the theme of a sentence closer to the sentence-initial position or as the rheme appearing in close proximity to the lexical verb in the finite position.[34]

4.2.2 Branching / Type: SOV / OV / SVO

Classified as a left-branching language, the (lexical) verbs in German often follow the object, always when the lexical verb is in the finite position (Müller 16). In simple ditransitive sentences without extraction, there are six options in the case of German verb-last clauses (e.g., weil/because), whereas in English there is only one (or two if a dative-shift, i.e. a prepositional object, is used): “[because] the woman gives the boy the book” (Müller 22). The position of adverbials[35] in German is rather free, as they can appear anywhere between the arguments and the verb (Müller 24-5).

The constituents of Russian sentence structure are even freer than in German (Bailyn 237). However, the branching classification and standard word order of Russian is controversial. Bailyn argues that it is SVO for transitive verbs; VS for intransitive verbs (238, 239). He adds nonetheless that there is a base word order (249).

4.2.3 Theme/Rheme

Scrambling in both Russian and German facilitates theme/rheme movement: Russian word order is not dictated by grammatical rules or minimalism, but rather permits “movement as a way of overtly establishing representations required by the discourse/functional component of grammar” (Bailyn 321, 324). One typical approach, which Bailyn calls Movement to the Far Left (MFL), involves placing a sentence constituent dislocated from the base SVO structure to the far left edge of the sentence to “fix” the theme, while the rest of the sentence is the rheme or informational focus (Bailyn 324). However, the options in Russian are even more extensive than what Müller showed (in section 4.2.2) for German: the rheme or the theme can be placed toward the finite position in a sentence with complex prepositional phrases or comparative functional heads.

4.3 Summary of syntax and relevance to belief updating

The SVO word order in German is indeed common in declarative sentences and after certain conjunctions and has been found to account for 79% of Russian sentences (Bailyn 249). Nonetheless, variants abound to an extent unknown in English. Accordingly, agents (readers, listeners) receiving a German or Russian sentence must constantly draw a connection between sentence- or clause-initial information and information toward the finite position. Peripheral constituents such as prepositional phrases and comparative functional heads frequently appear between the central constituents at or near both the initial and finite position. This greater diversity of syntax in German and Russian relative to overwhelmingly SVO syntax in English encodes divergent cognitive practices.

It can be expected that these three features of syntax, without consideration of further ones, influence how language users (through life-long repetition) update beliefs. In section five, I will argue that peripheral constituents such as comparative functional heads between subject-DPs and object-DPs or verb phrases will foster belief updating on peripheral networks, whereas noun-verb-noun patterns will move peripheral networks to the sideline, requiring either rejection or the activation of the central network.

5. The exits (in English, German and Russian): syntax as directing peripheral and central updating

We have passed the (reason-responsive) road signs on the Spinozan-Kantian endorsement/rejection of cognition objects (section one). We have checked the map for central and peripheral routes in cognition (section two) and found that the signposts for cognition (e.g., discounting cues) are analogous to operators in language (e.g., the negation operator) (section three). It is understood that an English, German or Russian route to the destination entails a different relationship to those (syntactic) signposts (section four). Now, on the theoretical road we have travelled, the syntactic signposts direct us to different exits and thus different (cognitive) routes.

Exit 1 (US English): belief updating is weighted toward bipolarity due to (American) English syntax and linguistic norms. The two poles are (i) effortful rejection or (ii) central network endorsement, with relatively little new information allocated to a peripheral network.

Exit 2 and 3 (German and Russian): the syntax processed, along with linguistic conventions in the respective cultures, formally replicate the cognitive representational architecture for dual reason-responsive belief updating on both a central and peripheral network.

5.1 Exit 1 (US English) – syntax directs toward central belief updating or rejection

The central belief updating of American cognition is a natural product of a mind that expects and processes its theme/rheme representations in concentrated blocks at or near the sentence-initial position. The core message of the belief is reliably conveyed in word and speech as an SVO or SP unit, and scrambling is limited (section four). An agent entertains the new belief without regarding prepositional phrases, comparative functional heads or other determiner phrases placed largely either before or after the SVO or SP unit. If the sleeper effect was determined to be strong for a discounting cue (negation operator) after the verb (section three), think about how qualifications via comparative information or prepositional phrases, which tend to be subsequent to the SVO/SP unit, will be forgotten over time.

The (new) belief becomes embedded in working memory as solely the unqualified SVO unit. Factors such as context or comparison are secondary for the agent because that information falls outside the scope of what must be taken up (SVO/SP). A mind processing information in this way is not used to accommodating context (prepositional phrases), contrast (comparative functional heads) or other sentence constituents. Working memory consists of SVO content. Endorsement or rejection depends on the relationship of the new belief to other beliefs already held. If in conflict, the belief will be rejected. If consistent, the central network will be updated. Exactly this result is what Festinger et al. (1959) found in subjects seeking to reduce cognitive dissonance (section two). And when voiced or acted on, the polarization (rejection vs. central updating) becomes even more pronounced as Lefebvre et al. (2024) showed with whatever statement of attitude an agent chose: it was reinforced by simply stating it, independently of echo chambers, filter bubbles or the absence thereof (Lefebvre et al. 2024: 7). Notably, and in contrast to German and Russian, peripheral networks are activated less for belief updating because language users in speech and working memory, are not used to processing sentence constituents besides the block of SVO or SP, especially when later asserting their belief in speech or action (Festinger et al. 1959; Lefebvre et al. 2024).

5.2 Exit 2 (German and Russian) – syntax guides toward peripheral belief updating

In German and Russian culture, a message is processed in syntax that is often scrambled, does not present the theme/rheme information in a bundle, but rather distributed with peripheral information (prepositional phrases, comparative functional heads, etc.) in between, and often requires deferment of belief comprehension until the end of the sentence (where either the subject- or object-DP or verb phrase is located). Automatic acceptance follows here, but, barring rejection, the subsequent endorsement readily occurs on either of the endorsement networks: central or peripheral. Both the distribution of theme/rheme information in the sentence containing the propositional content of the belief and the scrambling of communicators in that language will qualify endorsement due to the balance between the theme/rheme information related to the belief and directly or indirectly related peripheral information conveyed in the process of grasping the belief, especially contrast and context information. Relative to the (American) English message recipient who often receives such peripheral contrast and context information after the theme/rheme information (subject-predicate-object), the German or Russian message recipient must frequently parse such information between the subject-DP and verb phrase or subject-DP and object-DP or object-DP and verb phrase, between the theme and the rheme.

Where speakers and agents are constantly passing through peripheral syntactic elements on the way to central ones, and thereby constantly reinforcing the peripheral and central, this framework, unsurprisingly, shows up in the updating of beliefs both centrally and peripherally. A (new) belief provisionally accepted for later endorsement or rejection is rejected less because, even if it is in conflict with beliefs held in the central cognitive network, the (new) belief is readily assigned to a peripheral network of cognition. The syntax has laid the grounds for such updating.

Conclusion

What I have described may account for other cultural differences. One manifestation is how common it is for people with different beliefs to interact with each other in Germany and Russia, while opposing views in the United States drive us apart. Newspapers do not publish columns about what topics to discuss at a meal in order to avoid politics[36] because cognition is not limited to primarily central cognitive beliefs in German- and Russian-speaking contexts. Such interlocutors are also intimately familiar with peripheral beliefs.

Yet differences in belief updating centrally and peripherally, especially the possibility of silent activation of peripheral networks in English relative to their open activation, voiced and acted on, in German and Russian are borne out in both cultural studies and literature if one views central belief updating as a type of belief closure (authoritative), whereas peripheral updating is open and allows for contradictory beliefs to be held propositionally (that p and that ~p). More works needs to be pursued on this topic, but here are some preliminary observations.

Aligned with the argument here, studies of Russian society and Russian literature reveal conversation practices involving lengthy discussion with no closure (i.e., absence of belief updating centrally) and often with openly contradictory propositions (that p and that ~p); whereas such a practice within the larger framework will be absent in English-language contexts of speech and behavior that strive toward resolution (even if such contradiction is possible in thought rather than speech or action in English). Such differences are documented by Nancy Ries in Russian Talk: Culture and Conversation during Perestroika and by Margaret Paxson in Solovyovo – The Story of Memory in a Russian Village.

Another example arises in literary fiction. In Brothers Karamazov, Dostoevsky presents opposing worldviews in the believer Alyesha and the quasi-atheist Ivan as they discuss at length accepting god, but not his world, without any conclusion.[37] Parallels of such spoken non-closing discourse in English-language classics are hard to find, but not with respect to thought. In Persuasion, for example, Jane Austen’s narrator produces an entire novel on Anne Elliot’s conflicting beliefs about love and Mr. Wentworth; the same applies to Elinor in Sense and Sensibility. The representation of these characters as holding a belief that p and that ~p, presumably on divergent peripheral cognitive networks, is the plot of each novel. But what is notable? Neither Anne nor Elinor speak or act on their belief, even with those closest to them. They solely think or cognize it.

In closing now, I would like to address a couple of possible objections to the argument I have made in this paper.

Objection 1: the existence of fragmented cognition cannot depend fully on language use

If German and Russians can update beliefs on central or peripheral cognitive networks, all humans must possess a similar architecture, at least the disposition to harbor beliefs on both networks. So goes the first objection.

Indeed, it would be inaccurate to assume that the forming of competitive belief systems is largely reserved for SOV or dual-branching languages like German and Russian or ones lacking predictability due to word order scrambling. After all, Petty and Cacioppo developed their Elaboration Likelihood Model on the basis of experiments with Americans, and a similar thing can be said of Mandelbaum in the case of his analysis of racial bias and seating position (Mandelbaum 2016: 641). I have already pointed out that this analysis is not a zero-sum game where Americans lack the duality and our European counterparts contain it due to syntax and practice. The argument has been that Americans are more likely to reject or endorse via a central cognitive network, especially when they act or voice such beliefs. A person’s attitude can be swayed in the short term by new information. The non-biased person may act in a biased manner when presented with the majority information, but unless they speak or act as the $1 subjects or as the attitude-expressing agents in Lefebvre et al., it stands to reason that they will soon forget this peripheral information just as they forget, analogous to the discounting cue, information following the SVO unit close to the sentence-initial position.

Objection 2: if we accept this theoretical model, what is the practical relevance?

Short answer: we would pay more heed, especially in the United States, to marginalized silent groups. Why?

Long answer: due to English syntax and linguistic convention, it is non-acting, non-speaking belief updaters who, like the religious believers with fragmented beliefs discussed above, are better able to house divergent belief systems on central and peripheral cognitive networks because they refrain from expressing (voicing or acting on) peripheral beliefs and thus turning them into centralized beliefs. However, here I mean people who shy away from expressing their central beliefs as well, since such expression runs the risk we examined in section five (forgetting the peripheral beliefs due to dominance of central ones). If the potential central-conflicting belief warrants attention, but cannot yet be endorsed centrally (due to conflict), such a person will store and contemplate it, especially if it is desirable, on a peripheral network just as Germans and Russians (but, again, only because their general belief-silence prevents central beliefs from dominating over peripheral ones).

Indeed, it is perhaps in part this development that began to take place on a large scale in race relations ~60 years ago in the United States. The updating of biases on all sides was initially assigned to peripheral networks – desired, but given low credence in terms of sustainability. Nothing was said. The update was only thought. Time passed. More representations formed syntheses in the faculty of the imagination, were attached to categories and continued to be recognized (silently) on peripheral cognitive networks. Eventually, when voiced by some, these beliefs instantiated a fact that moved a peripheral belief to the central network. It was in this context of an updated central network rejecting racism through voice and action that parents like my own sent their European-American children to majority African- and Latino-American elementary schools in New York City (in the early 1980s) where they were fully accepted by the local majority with parents speaking and acting with likewise updated, now central beliefs.

Despite success stories, the polarization on many beliefs in American society continues to suggest the inhibition of a peripheral network in belief updating. Perhaps this is due to the premature voicing of opinions facilitated by online platforms due to the “enormous number of opportunities for the person to reinforce their position by simply stating it and collecting random, incidental feedback” independently of echo chambers and filter bubbles (Lefebvre et al. 2024: 7). Perhaps a silent response to a new belief is the recipe for fostering peripheral networks capable of balancing out the inevitable primacy of a central network and tempering belief-based tension in society. Perhaps it is some other facet of the non-English-speaking, non-American world that can help guide us toward a society of greater perceived equality not just between central and peripheral beliefs, but also between each other. What I have attempted to do here is show how differences in the syntax of language may account for the polarization we see in American minds today.

Version: December 2025.

Works cited

Amici, Federica et al. “The word order of languages predicts native speakers’ working memory.” Scientific Reports 9: 1124 (2019): 1-8.

Arnett, J.J. “The Neglected 95%: Why American Psychology Needs To Become Less American.” American Psychologist V. 63 (2008): 602-614.

Austen, Jane. Sense and Sensibility. Oxford, UK: Oxford University Press, 1998.

—. Persuasion. London, UK; New York, USA: Penguin Books, 1998.

Batson, C.D. “Rational processing or rationalization? The effect of disconfirming information on a stated religious belief.” Journal of Personality and Social Psychology 32 (1) (1975): 176-184.

Bendaña, Joseph, and Eric Mandelbaum. “The Fragmentation of Belief.” In The Fragmented Mind. Oxford University Press (2021): 78-107.

Benitez-Burraco, Antonio. “The cognitive science of language diversity: achievements and challenges.” Cognitive Processing 26 (2025): 719-736.

Bailyn, John Frederick. The Syntax of Russian. Cambridge, UK: Cambridge University Press, 2011.

Blasi, Damián E., Joseph Henrich, Evangelia Adamou, David Kemmerer, and Asifa Majid. “Over-reliance on English hinders cognitive science.” Trends in Cognitive Sciences. V. 26 No. 12 (2022): 1153-1170.

Dostoevsky, Fyodor. Brothers Karamazov. Trans. Michael R. Katz. New York, USA: Liveright Publishing Corporation, 2023.

Festinger, Leon, and James M. Carlsmith. “Cognitive Consequences of Forced Compliance.” The Journal of Abnormal and Social Psychology 58, no. 2 (1959): 203-10.

Fichtner, Edward G. English and German Syntax: A Contrastive Analysis on Generative-Tagmemic Principles. Munich, Germany: Wilhelm Fink Verlag, 1979.

Foos, Adrienne E., Kathleen Keeling, Debbie Keeling. “Redressing the Sleeper Effect: Evidence for the Favorable Persuasive Impact of Discounting Information Over Time in a Contemporary Advertising Context.” Journal of Advertising 50, No. 1 (2016): 19-25.

Kant, Immanuel. Critique of Pure Reason. Trans. Paul Guyer and Allen Wood. Cambridge, UK: Cambridge University Press, 1998.

Gendler, Tamar Szabó. Intuition, Imagination, and Philosophical Methodology. Oxford, UK: Oxford University Press, 2011.

Hovland, Carl and Walter Weiss. “The influence of source credibility on communication effectiveness.” Public Opinion Quarterly 15 (4) (1951): 635-650.

King, Tracy Holloway. Configuring topic and focus in Russian. Dissertation. Stanford University, 1993.

Lefebvre, Germain, Ophélia Deroy and Bahador Bahrami. “The roots of polarization in the individual reward system.” Proc. R. Soc. B 291: 20232011 (2024): 1-10.

Mandelbaum, Eric. “Thinking is Believing.” Inquiry: An Interdisciplinary Journal of Philosophy. V. 57:1 (2014): 55-96.

Mandelbaum, Eric. “Attitude, Inference, Association: On the Propositional Structure of Implicit Bias.” Noûs 50, no. 3 (2016): 629-58.

Mandelbaum, Eric. “Trouble with Bayesianism: An Introduction to the Psychological Immune System.” Mind & Language 34, no. 2 (2019): 141-57.

Müller, Stefan. German syntax: A constraint-based view. Berlin: Language Science Press, 2023.

Paxson, Margaret. Solovyovo – The Story of Memory in a Russian Village. Indianapolis, USA: Indiana University Press, 2005.

Petty, Richard, E. and John T. Cacioppo. “Personal Involvement as a Determinant of Argument-Based Persuasion.” Journal of Personality and Social Psychology V 41 (5) (1981): 847-855.

Petty, R.E. and D.T. Wegener. “The Elaboration Likelihood Model: Current Status and Controversies.” Dual Process Theories in Social Psychology. Eds. S. Chaiken and Y Trope. New York, USA: Guilford Press, 1999.

Pion, Griffen, Elliot Schwartz and Eric Mandelbaum. “Believe What We Think!: The Spinozan Theory of Mind.” The Oxford Handbook of the Cognitive Science of Belief. Eds. Van Leeuwen, Neil and Tania Lombrozo.

Pratkanis, Anthony R., Anthony G. Greenwald, Michael R. Leippe and Michael H Baumgardner. “In Search of Reliable Persuasion Effects: III. The Sleeper Effect Is Dead. Long Live the Sleeper Effect.” Journal of Personality and Social Psychology. Vol 54, No. 2 (1988): 203-218.

Pütz, Martin and Marjolijn Verspoor. “Introduction.” Explorations in Linguistic Relativity. Ed. Martin Pütz and Marjolijn H. Verspoor. John Benjamins Publishing Company, 2000: ix-xvi.

Ries, Nancy. Russian Talk: Culture and Conversation during Perestroika. Ithaca, NY: Cornell University Press, 1997.

Quine, Willard Van Orman. “Two Dogmas of Empiricism.” The Philosophical Review. Vol. 60, No. 1 (1951): 20-43.

Quine, Willard Van Orman. Word and Object. Cambridge, USA: The MIT Press, 1960.

Quine, Willard Van Orman. Ontological Relativity and Other Essays. New York: Columbia University Press, 1969.

Rehbein, Jochen. “Zur Wortstellung im komplexen deutschen Satz.” Deutsche Syntax: Ansichten und Aussichten. Ed. Ludger Hoffmann. Berlin, Boston: de Gruyter, 1992: 523-574.

Rydell, Robert J. and McConnell, Allen R. “Understanding Implicit and Explicit Attitude Change: A Systems of Reasoning Analysis.” Journal of Personality and Social Psychology. Vol. 91, No. 6 (2006): 995-1008.

Sharot, Tali, Max Rollwage, Cass R. Sunstein, and Stephen M. Fleming. “Why and When Beliefs Change.” Perspectives on Psychological Science 18, no. 1 (2023): 142-51.

Spinoza, Benedict. Spinoza Reader: The Ethics and Other Works. Edited and translated by Edwin Curley. Princeton, USA: Princeton University Press, 1994.

Thalmayer, Amber Gayle. “The Neglected 95% Revisited: Is American Psychology Becoming Less American?” American Psychologist V. 76 No. 1 (2021): 116-129.

Trabant, Jürgen. “How relativistic are Humboldt’s Weltansichten”? Explorations in Linguistic Relativity. Ed. Martin Pütz and Marjolijn H. Verspoor. John Benjamins Publishing Company, 2000: 25-45.

[1] The notable exception here is Amici et al., who will be referenced in support of the argument in this paper.

[2] The following clauses in German and English are formulated in what one would call conventional syntax. In Russian, there is no fixed word order, so anything is possible. See section four here for detailed remarks.

[3] For a clearly laid-out diagram of the respective models, see Mandelbaum 2014: 60 (Cartesian), 62 (Spinozan) or Pion et al: 3 (latter work is forthcoming, page number may be inaccurate).

[4] Intuition in the Kantian framework is not to be equated with intuition in Spinoza.

[5] See, inter alia, Kant A 97: A threefold synthesis, which is necessarily found in all cognition: the synthesis of the apprehension of the representations, as modifications of the mind in intuition; of the reproduction of them in the imagination; and of their recognition in the concept. (A 97)

[6] The synthesis of the manifold (representations of objects) takes place in the imagination (Kant A 78-79, A 97, A 102), facilitating the application of a concept (by the faculty of the understanding) to the representation provided by the intuition. If this did not happen, we would just encounter a haphazard, random stream of representations in our minds and be unable to process them; i.e., no combination, separation, categorization (application of Kantian category), etc. would be possible.

[7] See Kant A 62 and A 95-96, B 149 and B 165-6: in A 62, the representations/appearances in the intuition are given by the faculty of sensibility, which interacts with the empirical world; in A 95-6 and B 149, Kant explains why a priori concepts must themselves contain nothing empirical, but must belong to possible experience (A 95-6); only if appearances of sensible and empirical intuition are possible can a category be applied to a concept (B 149); finally and directly, Kant writes: “We cannot think any object except through categories; we cannot cognize any object that is thought except through intuitions that correspond to those concepts. Now all our intuitions are sensible, and this cognition, so far as its object is given, is empirical. Empirical cognition, however, is experience. Consequently, no a priori cognition is possible for us except solely of objects of possible experience.” (B 165-6)

[8] However, if a peripheral attitude change mechanism pairs the attitude object with positive (or negative) affective cues over many trials, this “old” conditioned attitude is likely to remain highly accessible and influence behavior, according to Petty and Wegener, especially if a recently changed attitude is not as accessible and there is insufficient time for contemplation (ibid 63).

[9] If Gendler’s aliefs have propositional content, this line of argumentation would also apply to the simultaneous mental state of believing, e.g., that the skywalk over the Grand Canyon is safe to walk on (that p), but alieving that it is dangerous based on the representational format (high platform), the affection (scary) and the behavior (stay away), i.e. that ~p. (see Gendler 267-8).

[10] I will ignore the issue of whether the auxiliary verb “does” should be disregarded. This is an important issue, but at the very least the findings here hold with respect to the verb “be.”

[11] See also Foos et al., 2016, who replicated the study with only slight modifications (11-day delayed assessment period) with similar findings: an increase in the favorability of attitudes for the message + post-discounting group after the 11-day period relative to the no-message group, while attitudes in the message + pre-discounting group did not increase compared to the no-message group (23).

[12] I intend to limit the discussion to common declarative sentences.

[13] The English “translation” of the Russian in each case here is obviously intended to convey representation of the content in the mind via the grammar of the original, most importantly, that the negation operator precedes the verb.

[14] The bracketed ellipses are intended to convey that in the past tense (Perfekt) a prepositional or object determiner phrase would be placed in the position of the brackets such that one might say that the negation operator precedes the main verb (even though it follows the auxiliary verb (haben)).

[15] The Russian language has a zero copula for the verb “be” in the present tense; in the past tense, however, the negation operator precedes the verb.

[16] In normal English: she did not do the work four days ago. The transposed “has” in brackets bears no relation to the present perfect form in English.

[17] See, e.g., experiment 17 and results (213-14).

[18] See remarks in conclusion for further repercussions.

[19] This table is by no means exhaustive. In some cases, for example the potential for scrambling in English, we cannot speak of a grammatical rule, but rather a convention widely adopted for understandability. In principle, prepositional phrases and other phrases or heads can be embedded, especially with commas for clarity, between sentence constituents (e.g., “he read, in the library, for the whole weekend, without any interruption, the book from his grandmother”), but for reasons of space and relevance to the belief updating argument, I will confine myself to convention (e.g., “he read the book from his grandmother in the library for the whole weekend without any interruptions”).

[20] See Müller (2023): 22 (scrambling); 24-5 (adverbials/prepositional phrases).

[21] See ibid (2023): 22 (scrambling); 24-5 (adverbials/prepositional phrases); also Bailyn (2011): 293.

[22] See Bailyn (2011): 237 f.

[23] See Amici et al. (2019): 1, 6-8.

[24] See Müller (2023): 25; Bailyn (2011): 293.

[25] Müller 16.

[26] See discussion below: Russian syntax for transitive verbs is generally categorized as SVO; for intransitive verbs as OV (Bailyn 238-9). However, any word order is possible. According to one study, “only” 79% of sentences fall under an SVO structure (Bailyn 249). In literary fiction, I would suggest this number is an order of magnitude lower.

[27] SVO for transitive verbs; VS for intransitive verbs (Bailyn 238, 239)

[28] See King 4.

[29] See Müller: 24-5, 31-2 (verbal complexes)

[30] As an aside, compare chart with specification of 7 tagmemes in respective slots for English (Fichtner: 65) with complete absence of such a chart for German (ibid: chapters five and six).

[31] See Müller (2023): 25; Bailyn (2011): 293.

[32] Prepositional phrases in the language of this paper and generative linguistics.

[33] See Müller (2023): 22 (scrambling); 24-5 (adverbials/prepositional phrases); also Bailyn (2011): 237f, 293.

[34] Rehbein provides a good example of this in German: “So hat sich schon vor einiger Zeit eine Gesellschaft konstituiert, deren ausschließliches Ziel es ist…” [literal translation: “so [was] some time ago a society established, whose sole aim it is…”] In the gloss, he writes, “So … schon vor einiger Zeit” constitutes the theme about which the new knowledge will be processed in a complex construction: additional new knowledge will be attributed to the rheme as subject phrase in the finite position (a society), which is expanded on in the dependent clause (Rehbein 552)

[35] Again, i.e., prepositional phrases.

[36] See NY Times, “Safe Topics to Discuss This Holiday Season”, Nov. 17, 2018: https://www.nytimes.com/2018/11/17/opinion/sunday/thanksgiving-family-political-discussion.html

[37] See Fyodor Dostoevsky, Brothers Karamazov, book five, chapters three to five, especially 277-314.