Elsevier

NeuroImage

Volume 141, 1 November 2016, Pages 341-349
NeuroImage

The effects of an action's “age-of-acquisition” on action-sentence processing,☆☆

https://doi.org/10.1016/j.neuroimage.2016.07.034Get rights and content

Highlights

  • Participants processed concrete/abstract aspects of late- and early-acquired actions.

  • Focusing on abstract aspects of actions resulted in “default mode network” activation.

  • Default network activation was higher when processing later-acquired actions.

  • This was the case also when participants focused on concrete aspects of an action.

  • We discuss implications for research on concrete and abstract cognition

Abstract

How does our brain allow us comprehend abstract/symbolic descriptions of human action? Whereas past research suggested that processing action language relies on sensorimotor brain regions, recent work suggests that sensorimotor activation depends on participants' task goals, such that focusing on abstract (vs. concrete) aspects of an action activates “default mode network” (rather than sensorimotor) regions. Following a Piagetian framework, we hypothesized that for actions acquired at an age wherein abstract/symbolic cognition is fully-developed, even when participants focus on the concrete aspects of an action, they should retrieve abstract-symbolic mental representations. In two studies, participants processed the concrete (i.e., “how”) and abstract (i.e., “why”) aspects of late-acquired and early-acquired actions. Consistent with previous research, focusing on the abstract (vs. concrete) aspects of an action resulted in greater activation in the “default mode network”. Importantly, the activation in these regions was higher when processing later-acquired (vs. earlier acquired) actions—also when participants' goal was to focus on the concrete aspects of the action. We discuss the implications of the current findings to research on the involvement of concrete representations in abstract cognition.

Introduction

Much of human activity revolves around communicating with other humans using the symbolic form of language; and a large portion of human linguistic-symbolic communication entails an exchange of information concerning the actions of other humans (Dunbar, 2004). This exchange of information concerning human actions plays an important role in shaping our behavior; for example, if you are told that “Donald stole the candy from the baby's mouth”, you may be less likely to vote for Donald for presidency. How does our brain allow us to comprehend and respond to such abstract and symbolic descriptions of human action? In recent years, two major alternatives have been suggested.

According to the “embodied cognition” framework (e.g., Barsalou, 1999, Niedenthal et al., 2005), humans' ability to understand action sentences is dependent on the retrieval of concrete sensory and motor representations. In a nutshell, this account suggests that in order to decipher the meaning of an abstract message such as “Donald stole the candy”, we must re-create the concrete perceptual and motor experiences associated with taking candy from a baby (for example, imagining grabbing the lollipop and overpowering the baby in a tug o' war; then gleefully running away, blonde hair scattered by the wind). Furthermore, embodiment theory suggests that humans are able to re-create these perceptual and motor states by re-activating neural systems that are involved in actual action and perception (e.g., Barsalou, 1999).

Indeed, much neuroscientific research has been brought forth to support these predictions. For example, research has shown that motor verbs (e.g., Hauk et al., 2004) and sentences (e.g., Tettamanti et al., 2005) activate the motor cortex in a somatotopic manner (i.e., hand-related words activate regions associated with hand movement, leg-related words activated regions associated with leg movement, and so forth). Furthermore, some neuropsychological evidence likewise suggested that the motor cortex may play a causal role in processing action-related semantics; for example, lesions to hand-related motor-primary motor cortex regions were shown to be associated with disrupted processing of action-related semantics using a stimuli set that included mostly hand-related sentences (Kemmerer et al., 2012).

However, it is important to note that the role of the motor system in action language processing remains under considerable debate. For example, a meta-analysis of action language processing did not find statistically significant activations in premotor and primary motor cortex regions (Watson et al., 2013). Furthermore, it has been shown that non-words can produce reliable activation in primary and premotor cortex regions, and that motor activation during action word processing may stem from the computation of ortho-phonological properties that are involved in deciphering grammatical classes (de Zubicaray et al., 2013). Finally, a study investigating the consequences of lesions to effector-specific motor regions did not find evidence for effector-specific semantic impairments (Arevalo et al., 2012).

In contrast to embodied cognition theory, traditional accounts of human cognition (e.g., Fodor, 1983) have long argued that the neural and cognitive systems that are responsible for abstract cognition—as the type involved in language comprehension—are functionally segregated from the systems that are responsible for processing concrete sensory and motor experiences. These views are supported by much research (e.g., Binder et al., 2009, Stephens et al., 2010) that shows that processing linguistic-symbolic meaning activate a set of regions that include the ventromedial prefrontal cortex (vmPFC), dorsomedial prefrontal cortex (dmPFC), lateral temporal cortex, temporo-parietal junction (TPJ), posterior cingulate cortex (PCC), and superior frontal gyrus, often collectively referred to as “the default mode network” (Raichle et al., 2001). This network of regions is also recruited when thinking about people's goals and mental states (e.g., Frith and Frith, 2006, Van Overwalle, 2009, Spunt et al., 2011), as well as in other abstract cognitive tasks such as considering future (e.g., Schacter et al., 2007, Gilead et al., 2013b) and counter-factual events (e.g., Van Hoeck et al., 2012). Importantly, this network has very little to no overlap with the neural systems that subserve sensorimotor processing. As such, these findings were taken to support the hypothesized classic distinction between abstract/symbolic cognition and concrete/sensory-motor cognition (e.g., Binder et al., 2009).

In an attempt to integrate the theorizing and research in “classic” and “embodied” theories of semantics, recent work has shown that when people focus on the concrete aspects of an action (e.g., think about “how Danny ate the bread”), they indeed activate sensory and motor regions; however, importantly, this is not the case when they focus on the more abstract aspects of the same action (e.g., think about “why Danny ate the bread”). Instead, when individuals focus on abstract aspects of the action they consistently and robustly activate the “default mode network” regions involved in social-cognitive and linguistic-symbolic processing (Spunt and Adolphs, 2014, Spunt et al., 2010, Spunt et al., 2011, Spunt and Lieberman, 2012, Spunt and Lieberman, 2013, Gilead et al., 2013a). Thus, according to this line of findings, the retrieval of sensory-motor representation may not be a necessary part of language comprehension, but rather, is dependent on the goals and processing mode of the perceiver.

In the current research we wish to build on this prior work (e.g., Spunt and Adolphs, 2014, Spunt et al., 2010, Spunt et al., 2011, Spunt and Lieberman, 2012, Spunt and Lieberman, 2013, Gilead et al., 2013a) and go one step forward, to suggest that for some actions, focusing on concrete aspects still calls upon abstract-symbolic mental representations. This is because many human actions are so intertwined with their symbolic significance, such that trying to think of the concrete aspects of these actions will nonetheless bring to mind abstract representations. Specifically, we suggest that one dimension that plays an important role in determining whether an action is represented abstractly or concretely is the age at which the person learned about this action.

Consider the case of a child learning to play drums at the age of six. In the terminology of the Piagetian framework (e.g., Inhelder and Piaget, 1958) this child is still in the “preoperational stage” of cognitive development: his/her symbolic thinking capacities have still not completely developed, and s/he is likely to focus on the concrete, sensory-motor properties of drum playing. However, if this child learns how to play the drums at the age of twelve (i.e., in Piagetian terms, a child that has passed the “concrete operational stage”, and can now perform abstract, “formal operations”), the acquisition of drum-playing behavior is likely to recruit more abstract cognition. For example, at this later age s/he is more likely to focus on the more abstract social significance of the action (it makes me look cool, it is part of my overarching goal to become a musician) and rely on symbolic representations to learn the skill (e.g., follow explicit rules and musical notation). Because of these major differences, it is possible that when one learns to play the drums at a later (rather than earlier) age, the abstract aspects of drum-playing become part and parcel of the mental representation of this action. Likewise, “drinking juice”, an action likely acquired at an early age, can perhaps be processed without bringing to mind any abstract and symbolic meaning. In contrast, “drinking beer”, is likely to be acquired at an age at which a person would be already aware of its symbolic meaning in the social world. As a consequence, one's knowledge concerning this action may be inseparable from more abstract cognitive processing.

Based on this reasoning, we suggest that even when people consider “how” later-acquired actions are performed, they will nonetheless tend to process the abstract significance of this action—as indexed by the elevated degree of activity in neural regions that are typically recruited in processing abstract aspects of actions (namely, the regions that are typically activated when participants think “why” an action is performed).

The notion according to which “age-of-acquisition” could play an important role in cognitive processing has been supported in research showing that early-acquired words are processed more quickly than later-acquired words (e.g., Turner et al., 1998), possibly due to their greater familiarity/frequency (Lewis et al., 2001) or greater imageability (Wilson et al., 2013; but see Izura et al., 2011). Furthermore, theories of the structure of semantic networks (e.g., Steyvers and Tenenbaum, 2005) suggest that our semantic system may be built in a hierarchical manner, such that later-acquired semantics are built on the scaffolds of earlier-acquired semantics. However, despite the recognition of the importance of age-of-acquisition in research into lexical and semantic processing, previous work has not investigated how the age of acquisition of an action affects the processing of action sentences—and whether the age of acquisition of an action moderates the degree to which action sentence processing relies on concrete vs. abstract mental representations.

In order to investigate these questions we conducted two studies in both of which participants focused on the concrete (i.e., “how”) vs. abstract (i.e., “why”) aspects of actions that they acquired relatively earlier or later in life. Consistent with previous work, we predicted that default network activity will be relatively higher when participants thought about “why” (vs. “how”) an action is performed. Furthermore, and most importantly, we predicted that default network activity will be higher for later-acquired (vs. earlier-acquired) actions, regardless of whether participants focused on “how” or “why” an action is performed.

Section snippets

Stimulus ratings

We created a list of 100 behaviors by complementing a verb with two different objects. The stimuli were constructed such that one of the verb-object pairings described a behavior that we suspected would be acquired at a relatively younger age (i.e., “scratching a mosquito's bite”) and the second verb-object pair described a behavior that we suspected would be acquired later in life (i.e., “scratching a lottery ticket”). We used extensive pre-testing to establish the supposed differences in age

Participants

Twenty-eight right-handed participants (17 females, average age 25.74, SD = 3.08, range 21–33 years) participated in the experiment. They were all native speakers of Hebrew, none had a history of neurological or psychiatric disorders, and all had normal or corrected-to normal vision. One participant was excluded from the final analysis due to failure to comply with task instructions. Participants gave written consent prior to taking part in the experiment. The study was approved by the

Neural activity associated with processing late-acquired behaviors (LA How > EA How) ∩ (LA Why > EA Why)

In order to find the neural correlates of processing late-acquired behaviors, we searched for the conjunction of activation associated with retrieving information regarding how late-acquired (LA), vs. early-acquired (EA) behaviors are performed, with the activations associated with retrieving information regarding why late-acquired (vs. early-acquired) behaviors are performed. Processing late-acquired behaviors resulted in activation of the regions collectively referred to as the “default-mode

Study 2

This study replicated Study 1 with a different set of stimuli, which were equated along all available dimensions, namely, positivity, social complexity, cognitive complexity, motor complexity, imageability, familiarity, frequency. This entailed that the age difference between late-acquired and early-acquired activities decreased from approximately 12 years in Study 1 to approximately 6 years in Study 2. Thus, in the present study the average age of acquisition for the late-acquired activities was

Discussion

In two studies, participants performed a semantic retrieval task wherein they processed the concrete (i.e., “how”) and abstract (i.e., “why”) aspects of late-acquired and early-acquired actions. Consistent with much previous research (e.g., Spunt et al., 2010, Spunt et al., 2011, Spunt and Lieberman, 2012, Gilead et al., 2014), focusing on the abstract, why (vs. concrete, how) aspects of an action resulted in greater activation in regions collectively referred to as the “default mode network” (

References (50)

  • R.P. Spunt et al.

    Validating the why/how contrast for functional MRI studies of theory of mind

    NeuroImage

    (2014)
  • K. Supekar et al.

    Development of functional and structural connectivity within the default mode network in young children

    NeuroImage

    (2010)
  • F. Van Overwalle et al.

    Understanding others' actions and goals by mirror and mentalizing systems: a meta-analysis

    NeuroImage

    (2009)
  • F. Adams et al.

    Modality and abstract concepts

    Behav. Brain Sci.

    (1999)
  • L.W. Barsalou

    Perceptual symbol systems

    Behav. Brain Sci.

    (1999)
  • J.R. Binder et al.

    Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies

    Cereb. Cortex

    (2009)
  • C.J. Brainerd

    Stage question in cognitive-developmental theory

    Behav. Brain Sci.

    (1978)
  • M. Brett et al.

    Region of interest analysis using the MarsBar toolbox for SPM 99

    NeuroImage

    (2002)
  • E.N. Cannon et al.

    Infants generate goal-based action predictions

    Dev. Sci.

    (2012)
  • A.M. Dale

    Optimal experimental design for event-related fMRI

    Hum. Brain Mapp.

    (1999)
  • G. de Zubicaray et al.

    Putting an “end” to the motor cortex representations of action words

    J. Cogn. Neurosci.

    (2013)
  • R.I. Dunbar

    Gossip in evolutionary perspective

    Rev. Gen. Psychol.

    (2004)
  • D.A. Fair et al.

    The maturing architecture of the brain's default network

    Proc. Natl. Acad. Sci. U. S. A.

    (2008)
  • E. Ferrer et al.

    Fluid reasoning and the developing brain

    Front. Neurosci.

    (2009)
  • J.A. Fodor

    The Modularity of Mind: An Essay on Faculty Psychology

    (1983)
  • Cited by (2)

    The authors declare no conflict of interest.

    ☆☆

    Supported by a grant from the Israel Science Foundation to Nira Liberman (grant No. 92/12) and by the I-CORE Program of the Planning and Budgeting Committee and the Israel Science Foundation (grant No. 51/11).

    View full text