Functional specificity for high-level linguistic processing in the human brain

“…selected the checkpoint with the best perplexity on the validation set and tested how well the model representations capture human neural (fMRI) responses to sentences in the language-selective network (Fedorenko et al., 2011) and human behavioral responses in a self-paced reading task (Figure 1D).”

“We also examined the specificity of this relationship by considering another network important for high-level cognition – the fronto-temporal language-selective network (Fedorenko et al., 2011).”

“blocking the replication of viruses [9,10].”

“Still, so far only two fMRI studies have directly compared neural activation to language and music in the same participants, and both have found little evidence for overlap.(81,82) This lack of overlap has been taken to support largely independent neural networks for linguistic and musical syntactic processing, howeve…”

“Each trial lasted 8 s (see Fedorenko et al., 2011, for details).”

“…Broca’s area in sentence processing have included a memory task (Blank, Balewski, Mahowald & Fedorenko, 2016; Fedorenko, Hsieh, Nieto-Castanon, Whitfield-Gabrieli & Kanwisher, 2010; Fedorenko et al. 2011, Fedorenko, Duncan & Kanwisher, 2012a; Fedorenko, Nieto-Castanon & Kanwisher, 2012b, 2012c).”

“Schematic fl ow of one block with meaningful sentence in CFS invisible experiment.”

“This task is described in detail in Fedorenko et al. (2010) and subsequent studies from the Fedorenko lab (e.g. Fedorenko et al. 2011, 2020; Blank et al. 2014; Blank, Balewski, et al. 2016a; Blank, Duff, et al. 2016b ; Pritchett et al. 2018; Paunov et al. 2019; Shain et al. 2020, among others;…”

“In the second session, participants completed easy and hard spatial working memory and arithmetic tasks (from Blank, Kanwisher, & Fedorenko, 2014; Fedorenko et al., 2011, 2013) designed to localise MDN and DMN.”

“Language comprehension involves an interplay between regions that are specifically important for linguistic processing, and multiple demand regions (Fedorenko et al., 2011, 2012, 2013; Fedorenko, 2014; Fedorenko and Thompson-Schill, 2014).”

“A shortcoming of group averaging of fMRI performance is that individual differences in activation patternswill not be detected (Fedorenko et al., 2011).”

“However, the exact locations, shapes, and sizes of language areas vary between individuals (Lipkin et al., 2022; Fedorenko et al., 2010, 2011).”

“Following [20], we pick four brain networks: language networks [39], visual networks [36], default mode network (DMN) [38], and multiple demand (MD) network.”

“Subjects had to keep track of four (Easy condition) or eight (Hard condition) locations in a 3 × 4 grid (Fedorenko et al., 2011).”

“Domain-specific networks encompass brain regions that are highly specialized for particular types of demand, for example, motor or language processes (Varley et al., 2005; Fedorenko et al., 2011; Monti et al., 2012).”

“The lack of the language system engagement during code comprehension adds to the body of work that demonstrates high input selectivity of these regions (Fedorenko et al., 2011; Jouravlev et al., 2019; Monti et al., 2009, 2012; Pritchett et al., 2018).”

“Given that Broca's area contains BA 44 and BA 45, and these are reported to have functional (e.g., Fedorenko et al., 2011; Friederici, 2011, 2017; Goucha & Friederici, 2015; Grodzinsky & Friederici, 2006; Hagoort, 2005; Hagoort & Indefrey, 2014; Matchin et al., 2017; Pallier et al., 2011; Tyler et…”

“a similar mechanism of selection and competitor inhibition, it is plausible that cognitive control—possibly language-specific (94)—plays a role in syntactic selection as well.”

“Consistent with human neuroscience (Fedorenko et al., 2011; Amal-ric and Dehaene, 2019; Ivanova et al., 2020), LLM language units are particularly selective for nat-ural language compared to arithmetic equations, C++ code, or random sequences of characters (all matching the number of tokens and…”

“In a functional magnetic resonance imaging (fMRI) study, Fedorenko, Behr, and Kanwisher (2011) found little or no response by functionally localized language regions to sequential mathematical tasks such as summing four consecutive numbers.”

“Similar to the present study, Fedorenko et al. (2011) also found language specific activation in LIFG during sentence comprehension, when compared with tasks engaging cognitive control and working memory load.”

“literal conditions across diverse phenomena with respect to three candidate networks: the language-selective network (Fedorenko et al., 2011), which supports language processing, the Theory of Mind (ToM) network (Saxe &”

“We also chose this task as it has been used previously as a language localizer to address similar questions of neural overlap [see Fedorenko et al., 2011].”

“…been used in this literature, including discrimination, anomaly/error detection, (Koelsch et al., 2005; Carrus et al., 2013), grammatical acceptability (Patel et al., 1998a; Patel, 2008), final-word lexical decision (Hoch et al., 2011), and memory/comprehension tasks (Fedorenko et al., 2009, 2011).”

“In a recent study, Federenko, Behr, and Kanwisher (2011) used a language localizer to map out language regions on an individual subject basis.”

“To date, the sentences and word list paradigm with written words has mainly been used in hemodynamic studies (Fedorenko et al., 2011; Friederici et al., 2000; Humphries et al., 2005, 2006; Jobard et al., 2007; Snijders et al., 2009; Stowe et al., 1998; Vandenberghe et al., 2002; Xu et al., 2005).”

“The superior temporal cortex is known to be involved in a broad range of auditory tasks, including simple tone discrimination (Daikhin and Ahissar, 2015), language (Fedorenko et al., 2010) and music (Fedorenko et al., 2011), and even social tasks (e.g. Deen et al., 2015).”

“We leveraged our knowledge about the core language network – a set of frontal and temporal brain areas that selectively support high-level language processing (Fedorenko et al., 2011), including lexico-semantic and combinatorial processing (Fedorenko et al.”

“The language network robustly bears out this prediction (e.g., Fedorenko et al., 2011).”

“There are also studies, however, arguing for domain-specific sequence processing mechanisms, especially for the processing of language (Fedorenko et al., 2011).”

“It is possible that as the phonological analysis becomes necessary for higher-level language processing, the acoustic analysis of speech sounds supported by the right hemisphere might no longer be optimal (Blumstein et al., 2005; Hickok and Poeppel, 2007; Fedorenko et al., 2011; Friederici, 2011; Myers, 2014).”

“All three of these regions have been implicated in combinatorial linguistic processing by a substantial number of studies (Bottini et al., 1994; Humphries et al., 2001; Vandenberghe et al., 2002; Mar, 2004; Lau et al., 2008; Fedorenko et al., 2011; Pallier et al., 2011), and thus these results provide a basis against…”

“To test whether the brain-behavior relationship we observed is selective to the MD network, we considered another large-scale neural network: the fronto-temporal languageselective network in the left hemisphere (Fedorenko et al., 2011) (Figure 5A).”

“…compared to various non-linguistic inputs and tasks, such as music perception (Fedorenko et al., 2012; Chen et al., 2023) or arithmetic computation (Fedorenko et al., 2011; Monti et al., 2012) (for review, see (Fedorenko et al., 2024)) and the language network only shows weak responses when…”

“While in healthy individuals cognitive control and language networks show dissociated patterns of activation during simple language tasks (Fedorenko et al., 2011), multiple studies have suggested that bilateral domain-general cognitive control networks are engaged alongside the language network in…”

“Thus, the semantic localizer essentially replicated the previously reported language network (Fedorenko et al., 2010, 2011).”

“In particular, clear separation exists between the language-processing system 7 and both (1) lower-level perceptual areas and (2) higher-level cognitive areas (see ref.”

“, 2020) generated from experimental data via a watershed method (Fedorenko et al., 2011).”

“, 2020)—in core linguistic processes like lexical access and semantic/syntactic composition, showing that such computations are instead carried out within the language-specific cortical areas (Fedorenko et al., 2011).”

“The hard > easy contrast has been found to robustly activate regions of the MD network (Fedorenko et al., 2011, 2013).”

“For instance, it was noted that, even in group-averaged estimates, which blur over individual differences ( 50 – 53 ), there are hints of substructure within the canonical DN ( 43 , 54 – 61 ); ToM tasks tend to activate a more anterior region of the inferior parietal lobe at or near the…”

“Language function is supported by a domain-specific, left-lateralized, frontotemporal language network (Fedorenko et al., 2010, 2011; Ojemann, 1991; Vigneau et al., 2006), that is selective for the semantic and syntactic properties of language (Fedorenko et al., 2010).”

“Although the LIFG has long been associated with syntactic function in particular (e.g., Fedorenko et al., 2011; Stromswold et al., 1996), several studies find that the LIFG is engaged by the presence of representational conflict, even when syntactic processing demands do not vary.”

“Biologically valid MRI findings typically form clusters of voxels within the brain,[24] because the adjoining and interconnected cellular elements in tissues that support a given information processing function are considerably larger than a voxel[34, 35].”

“This cognitive process is mediated by a network of cortical areas in lateral temporal and lateral prefrontal cortices (Friederici, 2011), which are collectively involved in language comprehension (Fedorenko et al., 2011) and language production (Sahin et al., 2009).”

“In NT individuals, math and language draw on distinct networks (e.g., Varley et al., 2005; Fedorenko et al., 2011; Monti et al., 2012; Amalric & Dehaene, 2019) but i) math processing is generally left-lateralized (Monti et al.”

“The sentence > list contrast in comprehension occasionally activates Broca’s area (Bedny et al., 2011; Fedorenko et al., 2011; Pallier et al., 2011), but these effects are much less consistent than for the ATL (Rogalsky and Hickok, 2011).”

“In general, this is in line with the previously suggested independence of prefrontal argument – order-related brain activity from working memory processes during sentence processing (Caplan et al., 2000; Fedorenko et al., 2011).”

“This result argues against ubiquitous between-group differences in information processing, and is in line with prior work that has suggested that the language network is functionally distinct from other high-level large-scale networks (e.g., Blank et al., 2014; Fedorenko et al., 2011; Fedorenko & Varley, 2016; Minerof…”

“Specifically, superior-posterior parts of the left IFG were found to be shared between domain-specific (language) and domain-general (verbal working memory and cognitive control) functions (Fedorenko et al., 2011).”

“…claims for a clear distinction between brain regions exclusively dedicated to language versus other cognitive processes (Chen et al., 2023; Fedorenko et al., 2011; Fedorenko & Blank, 2020; Friederici, 2020) and for the existence of specific and separate neural populations for speech, music, and…”

“15 the speech and language network (Ayyash et al., 2021; Fedorenko et al., 2011; Hickok & Poeppel, 2007; Rauschecker & Scott, 2009), and ventral sensorimotor cortex and insula are regions that are clearly implicated in speech motor control (Bouchard et al.”

“Each trial lasted 8 s (see (Fedorenko et al., 2011) for the timing details).”

“This aligns with the general view that cognitive control at the higher levels of language processing draws on specific neural resources distinct from those engaged in other cognitive domains (Fedorenko et al. 2011; Mineroff et al. 2018).”