Gilead et al. lament that theories of abstract cognition have been left unintegrated in part because of a lack of terms of discourse common across branches of the cognitive sciences, or even between social and biological aspects of psychology. There is indeed some irony in this all too lowly construed approach to the cognitive science of construal and abstraction, distinct threads of which have been appearing in the history of cognitive science for at least the past seven decades. Our story begins with Witkin's (Witkin & Asch Reference Witkin and Asch1948; Witkin et al. Reference Witkin, Dyk, Fattuson, Goodenough and Karp1962) notion of field dependence in perception and psychophysics, and its subsequent relationship to gestalt-orientated cognition and to social affiliation and perspective-taking (Witkin & Goodenough Reference Witkin and Goodenough1977). This same idea of a concrete–abstract representational axis cutting across perceptual and social aspects of cognition was recapitulated by Frith (Reference Frith1989) and Frith and Happé (Reference Frith and Happé1994) as central coherence in describing both autism's decontextualised detail-orientated perceptual stance and its likewise decontextualised egocentric social perspective. Around the same time the idea was introduced to social psychology by Trope (Reference Trope1989) first as an account of dispositional trait versus situational state explanations of others’ behaviour, then extended to effects of temporal and other psychological distances on what Trope et al. had come to call perceptual construal (Trope & Liberman Reference Trope and Liberman2003), the term adopted in the rest of this commentary.

The syndrome of autism, along with its dimensional extension to individual differences in autistic (or what Witkin called field-independent) traits, exemplifies this association between construal and psychological distance: Spatial, temporal, social, and hypothetical distances resurface as autistic differences in mapping between allocentric and egocentric space (Conson et al. Reference Conson, Mazzarella, Esposito, Grossi, Marino, Massagli and Frolli2015; Frith & de Vignemont Reference Frith and de Vignemont2005; Hamilton et al. Reference Hamilton, Brindley and Frith2009; Pearson et al. Reference Pearson, Marsh, Hamilton and Ropar2014; Ring et al. Reference Ring, Gaigg, Altgassen, Barr and Bowler2018), impulsivity and executive disinhibition (Hill Reference Hill2004), social perspective-taking and other aspects of cognitive empathy (Baron-Cohen Reference Baron-Cohen1995), and repetitive-behavioural aversion to unpredictability and change (Gomot & Wicker Reference Gomot and Wicker2012). Gilead et al. relate the distinction between raw perceptual observations and elaborated cognitive models (abstracta) to the contrast between detail-orientated, first-person simulation and abstract, allocentric theory in predicting the behaviour of the world; impairment in prediction when constraints are underspecified, dynamic, or real-time – as is the case in social cognition – has been identified time (Courchesne & Allen Reference Courchesne and Allen1997) and again (Sinha et al. Reference Sinha, Kjelgaard, Gandhi, Tsourides, Cardinaux, Pantazis, Diamond and Held2014; Van de Cruys et al. Reference Van de Cruys, Evers, Van der Hallen, Van Eylen, Boets, de-Wit and Wagemans2014) as a unifying feature of autism which may drive the co-occurrence of anxiety and rituals, perceptual dysmodulation, visuomotor deficits, slowed orienting of attention, and undifferentiated processing of stimuli regardless of task-relevance. Because autistic predictions tend to be founded more on iconic, concrete perceptual data rather than on abstracta, they evoke many violations of expectation in instances where observations would match the broad strokes of an abstract model yet fail to match these minutiae (Van de Cruys et al. Reference Van de Cruys, Evers, Van der Hallen, Van Eylen, Boets, de-Wit and Wagemans2014). This hyper-reliance on iconic representations produces a style of cognitive inference by bricolage, that is, by effortful construction and maintenance of complex representations and ideas bottom-up from the underlying details and instances (Belmonte Reference Belmonte2008a), which are preserved in lieu of abstracta (Belmonte Reference Belmonte and Osteen2008b). This flattening of Gilead et al.'s hierarchy of abstracta implements a cognitive style adroit at recognising relationships amongst numerous, low-construal percepts, described by Baron-Cohen et al. (Reference Baron-Cohen, Ashwin, Ashwin, Tavassoli and Chakrabarti2009) as “systemising.” Although it can confer superiority at detail-orientated disciplines, this systemising style imposes such a great cognitive representational load that it cannot scale. Because predictions based on inappropriately detailed cognitive models frequently evoke mismatches with observations, and such errors of accidental detail are not differentiated from errors of essence (Van de Cruys et al. Reference Van de Cruys, Evers, Van der Hallen, Van Eylen, Boets, de-Wit and Wagemans2014), the world amounts to a constant chaos of Heraclitean flow in which one's expectations are always and inexplicably wrong, sabotaging social and other domains of reward and thus impairing learning and development. It's no surprise, then, that Gilead et al.'s hierarchy of representational qualities – from concrete, iconic, modality-specific impressions, through multimodal convergences (Brandwein et al. Reference Brandwein, Foxe, Butler, Russo, Altschuler, Gomes and Molholm2013; Reference Brandwein, Foxe, Butler, Frey, Bates, Shulman and Molholm2015; Ostrolenk et al. Reference Ostrolenk, Bao, Mottron, Collignon and Bertone2019), to socially communicated, categorical abstractions (Beker et al. Reference Beker, Foxe and Molholm2018; Feldman et al. Reference Feldman, Dunham, Cassidy, Wallace, Liu and Woynaroski2018; Smith et al. Reference Smith, Zhang and Bennetto2017; Stevenson et al. Reference Stevenson, Baum, Segers, Ferber, Barense and Wallace2017) – is exactly the hierarchy of perceptual and representational abnormality in autism.

All this evidence shows Gilead et al.'s ontology of abstraction and prediction to be consistent with historical concepts and findings, and with what we know about autism, its prime test case. But retrospection is the game of Monday-morning quarterbacks – what of prospective predictions, and experiments yet to be performed? Drawing together all these strands can relate behavioural and neural aspects of prediction and abstraction, psychological distance and construal, with corollary implications for cultural and sex differences in cognition: Gilead et al. speculatively peg the default-mode network as the home of their cognitive abstracta, although the true locus may lie rather in this network's interactions with other control networks. The default-mode network is constitutively active in autism (Kennedy et al. Reference Kennedy, Redcay and Courchesne2006), perhaps reflecting constant and largely fruitless attempts at predictive modelling (Raichle Reference Raichle2015) of accidental detail, associated with low-construal impulsive action (Shannon et al. Reference Shannon, Raichle, Snyder, Fair, Mills, Zhang, Bache, Calhoun, Nigg, Nagel, Stevens and Kiehl2011) and anxious affect (Simpson et al. Reference Simpson, Drevets, Snyder, Gusnard and Raichle2001a; Reference Simpson, Snyder, Gusnard and Raichle2001b).

The female advantage in default-mode network deactivation in reward contexts (Dumais et al. Reference Dumais, Chernyak, Nickerson and Janes2018) seems consistent with autism's association with male-typical cognition (Baron-Cohen et al. Reference Baron-Cohen, Knickmeyer and Belmonte2005), linking construal to cognitive sex differences. And Witkin (Reference Witkin1979) himself noted that construal variations can be a function of culture; indeed individualistic cultures are associated with a more systemising bias (Markus & Kitayama Reference Markus and Kitayama1991; Nisbett & Masuda Reference Nisbett and Masuda2003) and collectivistic cultures with higher construal (Boduroglu et al. Reference Boduroglu, Shah and Nisbett2009; Masuda & Nisbett Reference Masuda and Nisbett2006). One might predict, then, associations of individual trait construal level (a.k.a. autistic traits, field dependence), situational state construal level, sex and/or gender, and individualistic/collectivistic culture with the frequency and/or duration of dynamic coupling of default-mode with attentional and executive control networks (Ryali et al. Reference Ryali, Supekar, Chen, Kochalka, Cai, Nicholas, Padmanabhan and Menon2016). The degree of network coupling would reflect individual and situational differences in the bias and range of model-driven feedback versus environmentally bound feedforward cognitive control of perception, action, and affect, and would be measurable with fMRI, or perhaps EEG/MEG (Kitzbichler et al. Reference Kitzbichler, Khan, Ganesan, Vangel, Herbert, Hämäläinen and Kenet2015). Such a study would afford an opportunity to reconstrue (as it were!) as a neurophysiological variable the diversity with which individual humans walk the tightrope between Aristotelian category and Heraclitean instance, between Lacan's (Reference Lacan1966) le symbole and la chose.