Brain and Cognition 135 (2019) 103582

Contents lists available at ScienceDirect

Brain and Cognition

journal homepage: www.elsevier.com/locate/b&c

Tool and object affordances: An ecological eye-tracking study T

Giovanni Federico , Maria A. Brandimonte
⁎

Laboratory of Experimental Psychology, Suor Orsola Benincasa University, Naples, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: In the present eye-tracking study, we analysed the visuo-spatial attentional patterns of participants looking at 3D
Affordances images depicting single tools and object-tool pairs. The object-tool pairs could be thematically consistent, the-
Object-tool pairs matically inconsistent or spatially inconsistent. During the first 500 ms of visual exploration, tools were fixated
Eye tracking longer on their functional area in all experimental conditions. However, extending the time-window of analysis
Reasoning-based approach
to 1750 ms, the visual scene was encoded in a faster and more suited-for-action way in the thematically con-
Action reappraisal
sistent condition (e.g., hammer-nail). Most important, the visual exploration of the thematically consistent pairs
focused on the manipulation area of the tool (e.g., the handle of the hammer) more than on its functional area
(e.g., the head of the hammer). Finally, when single tools were shown and the entire time-window of analysis
was considered (1750 ms), fixation focused on the tool’s manipulation area. These results are discussed within
the reasoning-based framework of tool use. They highlight the relative role of the visuo-perceptual context in
affordance perception and suggest a novel interpretation of the cognitive mechanisms underlying the processing
of tools and object-tool pairs in terms of action reappraisal (i.e., a re-functionalization process when the action
possibility is mined by the visuo-perceptual context).

1. Introduction in affordance perception. In order to qualify and quantify affordance

perception, we chose to analyse gaze behaviour. More precisely, we
The use of tools and the capability to properly recognise them re- used eye-tracking techniques to investigate the allocation of visuo-
present a fundamental facet of the human intrinsic ability to interact spatial attention to different areas of a visual scene representing single
with the environment. A large amount of evidence has been produced tools or object-tool pairs. When appropriate, we used different notions
by research over the last forty years in the domain of tool affordances, of affordance perception proposed in the literature. Nonetheless, those
particularly when considering the most recent studies related to the notions, which were expressed through different terms, appeared to be
embodied cognition of perceptual and semantic processing of functional conceptually similar in the way they referred to different degrees of
and motor properties of objects (see Osiurak, Rossetti, & Badets, 2017 affordance perception evoked by the emerging characteristics of the
for a recent review). visuo-perceptual context (e.g., the relationship between objects com-
From its initial delineation, the concept of “affordance” (Gibson, posing an object-tool pair).
1977, 1979) acquired over time a magnitude of meanings, especially Analogously, the word “tool” may have different connotations.
when considered in the field of tool use. Different notions of affordance When considered in the domain of the embodied cognition and for the
generated sometimes confusion in the literature. In line with the defi- purposes of the present study, the notion of tool is referred to any
nition recently proposed by Osiurak et al. (2017), in the present study, handheld physical implementation that amplifies the user’s sensor-
we considered an affordance as “an animal-relative, biomechanical imotor capabilities; in this way, it is possible to use the word “object” in
property specifying an action possibility within a body/hand-centered order to refer to the plausible recipient of an action (Osiurak, Jarry, &
frame of reference” (Osiurak et al., 2017, p. 410). Accordingly, an af- Le Gall, 2010). Tools are a class of objects with intrinsic action and
fordance is “a description of this [action] possibility at a physical, but motor features, as even the passive observation of a tool engages the
not at a neurocognitive level” (Osiurak et al., 2017, p. 410). Therefore, activation of the same brain areas that are typically involved in its use
at a neurocognitive level, the issue remains to understand how an af- (Chao & Martin, 2000; Johnson-Frey, 2004; Króliczak & Frey, 2009).
fordance can be perceived. Our investigation was conducted with the The vision of a tool activates the motor system, so that tools, and ma-
aim to further our understanding of the cognitive mechanisms involved nipulable objects in general, are processed in terms of their affordances

⁎
Corresponding author at: Laboratory of Experimental Psychology, Suor Orsola Benincasa University, Via Suor Orsola 10, 80135 Naples, Italy.
E-mail address: [email protected] (G. Federico).

https://doi.org/10.1016/j.bandc.2019.103582
Received 14 April 2019; Received in revised form 27 May 2019; Accepted 22 June 2019
Available online 27 June 2019
0278-2626/ © 2019 Elsevier Inc. All rights reserved.
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

and motor qualities – e.g., “manipulability” or “graspability” – along- of space in affordance perception has also been extensively studied. For
side their visual attributes (for a review: Thill, Caligiore, Borghi, instance, it has been suggested that there is a motor activation only
Ziemke, & Baldassarre, 2013, but see also: Tucker & Ellis, 2001). when objects are presented within the peri-personal space of partici-
However, it is quite uncommon to observe a tool in an uncontested pants (Costantini, Ambrosini, Tieri, Sinigaglia, & Committeri, 2010). It
manner, seeing it alone, isolated or out of a visual scene with contextual is also well known that unreachable objects may not produce motor
references. In other words, in the real world, tools are part of a larger activation at all. An EEG study showed that the mu-rhythm desyn-
scene that includes – beside its physical, geometric and spatial prop- chronization – which is supposed to reflect motor preparation and ap-
erties – the set of contextual and spatial relations with other objects pears during motor imagery – was present only when objects were
(Mizelle & Wheaton, 2010). perceived as reachable (Pfurtscheller & Da Silva, 1999; Ritter,
While there is a general consensus on the automaticity of affordance Moosmann, & Villringer, 2009; Wamain, Gabrielli, & Coello, 2016). It
perception as related to single tools, only a small number of studies appears that unreachable objects may only evoke “functional affor-
have explored the mechanisms and the processes underlying the af- dances” (Buxbaum & Kalénine, 2010), whereas correctly located object
fordance perception for object-tool pairs. These pairs were mainly used pairs facilitate their identification (Roberts & Humphreys, 2011a,
to experimentally manipulate the context of the visual scene using pre- 2011b). These results shed new light on the way object pairs are pro-
ordered categories of relation between objects (e.g., thematic, func- cessed according to the context of the visual scene and stimulated us to
tional, spatial relations, etc.). The interest for paired-objects action- consider the role of the spatial relations between objects in terms of the
based effects has increased after neuropsychological studies showing kinematic properties exhibited by the visual scene. Thus, a hammer-nail
that patients with visual extinction – a phenomenon associated with pair, although thematically consistent because the objects composing
parietal damages in which patients fail to report the contra-lesional the pair are linked together by a functional relation, might not elicit any
stimulus when two stimuli are simultaneously presented – may benefit action possibility if the nail is placed in a position such that it cannot be
from action relations between objects, i.e., when the two object stimuli hit by the hammer, or if the hammer points towards the opposite di-
are correctly positioned for action (e.g., hammer and nail in adequate rection of the nail.
orientation for co-action; Riddoch, Humphreys, Heslop, & Castermans, In addition to their motor properties, tools – as a part of the class of
2002; Riddoch et al., 2006). objects – inherit some theoretical assumptions related to their semantic
Tools are always within a perceptual context, so that the context can properties. Indeed, the domain of object semantics includes two distinct
modulate affordance perception. Many studies with healthy partici- kinds of knowledge: the function knowledge and the manipulation
pants investigated, by using different paradigms, the facilitation effects knowledge. Function knowledge refers to the goal attainable by using
emerging from the arrangement of pairs of objects according to dif- an object. It is a kind of semantic knowledge that links tools with their
ferent kinds of relationships between the objects (Borghi, Flumini, common purposes and their associated actions, describing their proto-
Natraj, & Wheaton, 2012; Green & Hummel, 2006; McNair & Harris, typical use, e.g., a hammer is used to pound a nail (Collette, Bonnotte,
2014; Roberts & Humphreys, 2011a, 2011b; Yoon, Humphreys, & Jacquemont, Kalénine, & Bartolo, 2016; Goldenberg & Randerath,
Riddoch, 2010). Therefore, it is reasonable that a bottle generates 2015; Goldenberg & Spatt, 2009). Manipulation knowledge, instead,
higher “action readiness” when it is seen next to a glass rather than a refers to the gestures that one has to execute in order to appropriately
cap, as the glass increases the motor properties of the bottle, whereas use an object. It is a kind of knowledge that mediates the proper
the cap does not afford a functional use of the bottle. Thus, when an handling of tools, implying a motor-based simulation. For instance, a
object-tool pair is presented in the visual scene, each object of the pair saw is typically held with a power grip (Ellis & Tucker, 2000); then, it is
does not automatically evoke just the action possibilities it usually af- pressed on the wood with an oscillating movement that goes back and
fords when presented alone, but also those related to the common or forth. In other words, manipulation knowledge contains the specific
uncommon use of the tool with the specific object of the pair (e.g., a motor representations that underlie the use of tools, including the skills
knife near a screw may suggest “screwing” rather than the prototypical that are supposed to be involved in the selection of the appropriate
“cutting” action possibility; Humphreys et al., 2010; Yoon et al., 2010; gestures (Buxbaum, 2001; Ellis & Tucker, 2000). Indirect suggestions
De Stefani, Innocenti, Bernardi, Campione, & Gentilucci, 2012; Thill about the relative role of function vs. manipulation knowledge in object
et al., 2013). semantics may also be found in the literature on “thematic relations”,
De Stefani et al. (2012), in an ecological setting, showed that when a which concern objects that play a complementary role in a given sce-
thematically consistent object-tool pair is presented (a full bottle and an nario and that can either be part of the same superordinate category
empty glass) a “working affordance” emerges. With this term, the au- (e.g., hammer-nail) or not (e.g., spoon-yogurt; Estes, Golonka, & Jones,
thors refer to a representation of a specific sequence of actions in which 2011; Mounoud, Duscherer, Moy, & Perraudin, 2007; Perraudin &
a tool and a functionally compatible object interact. Yoon et al. (2010) Mounoud, 2009; Kalénine, Mirman, Middleton, & Buxbaum, 2012;
reached the same conclusion after presenting participants with virtual Pluciennicka, Wamain, Coello, & Kalénine, 2016). While functional
images of thematically consistent object-tool pairs. Specifically, they relations refer to associations between objects of the same kind (e.g.,
showed that when right-handed participants decide if two objects are hammer and saw), thematic relations rely on contextual links between
typically used together, speeded classification responses emerge when objects that are commonly used together, but that are not of the same
the two objects are positioned in a standard right-handed co-location. kind (e.g., hammer and nail; Kalénine et al., 2009).
Analogously, in order to investigate if affordance perception was Behavioural studies with healthy adult participants have provided
modulated by the perceptual context, Borghi et al. (2012) used object- evidence of a faster processing of function knowledge of an object ra-
tool pairs in a behavioural task. They reported faster responses when ther than of its manipulation knowledge (Garcea & Mahon, 2012). A
the relationship that tied the objects of the pair was a functional rather double dissociation between function and manipulation knowledge was
than a spatial relation or a no type relation (objects that do not share reported in patients with left brain damage (Buxbaum & Saffran, 2002).
any semantic or spatial link). Intriguingly, a recent study has revealed, Neuroimaging evidence about function and manipulation knowledge
using a semantic-free paradigm, that “identity” or “functional” knowl- are quite controversial: according to Boronat et al. (2005) both kinds of
edge of objects is not a prerequisite for the affordance perception of knowledge recruit the same frontoparietal regions of the visuomotor
object-tool pairs, as “it is possible to perceive affordances between system with a greater activity for manipulation, whereas Canessa et al.
objects by processing action-related structures across paired-object (2007) concluded that there are contrasting activations in the parietal
scenarios directly, when semantic knowledge is absent” (Xu & Heinke, and the inferior temporal cortex for the manipulation and function
2017, p. 15). knowledge. The authors reported significantly higher activation for
Besides thematic and functional properties of objects, the influence manipulation than function tasks in the left inferior parietal lobe, the

2
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

premotor cortex and the inferior parietal sulcus, whereas the inferior actions that can be realistically executed with them.
temporal lobe was significantly more activated for function-related Tool observation activates the left motor-related brain areas (the left
tasks. Similarly, recent results from experiments using semantic premotor and the posterior parietal cortex) alongside with the posterior
priming and eye-tracking paradigms are controversial as to the relative temporal areas that are involved in the visual motion processing. The
role of manipulation vs. function knowledge in object identification activation of the dorsal network is supposed to be the neural substrate
(Kalénine et al., 2012; Labeye, Oker, Badard, & Versace, 2008; Myung, of affordance perception, as it reflects the activation of motor routines
Blumstein, & Sedivy, 2006; Schreuder, d'Arcais, & Glazenborg, 1984; for the possible interactions with tools (Creem-Regehr & Lee, 2005;
Wamain, Pluciennicka, & Kalénine, 2015; Yee, Huffstetler, & Beauchamp & Martin, 2007; Grèzes & Decety, 2002; Jeannerod, Arbib,
Thompson-Schill, 2011). Overall, it seems that both function and ma- Rizzolatti, & Sakata, 1995). It appears that there is a specific neuroa-
nipulation knowledge may be implicitly activated during object visual natomical correlation that supports the existence of a specific paired-
processing. However, it remains unclear how these two types of object affordance effect when tools and objects are observed. Recent
knowledge interact, as well as the impact of the visuo-perceptual con- neuroimaging evidence has shown a joint contribution of dorsal (sup-
text in modulating either type of knowledge. plementary motor area) and ventral (left occipital cortex) streams to the
Within the perspective according to which humans use tools to solve paired-object affordance effect. In particular, there is an increased left
problems (e.g., Beck, Apperly, Chappell, Guthrie, & Cutting, 2011; occipital cortex activation when objects are correctly positioned for
Mounoud, 1996), recent lines of research – essentially derived from the action, while the anterior regions of the dorsal stream are activated
neuropsychological domain – challenged the long-established “manip- when the task requires an action decision, but objects are not in a
ulation-based” assumption (e.g., Buxbaum, 2001) for which perceived correct position for action (Roux-Sibilon, Kalénine, Pichat, & Peyrin,
affordances are motorically activated by the mere observation of tools 2018).
(see Osiurak & Badets, 2016 for an extended review). Hence, when In the field of visual cognition, very few studies analysed by eye
seeing a tool, a “reasoning-based” agent may not automatically activate tracking the visual encoding of tools (Ambrosini & Costantini, 2017;
manipulation knowledge (i.e., sensorimotor information about how to Myachykov, Ellis, Cangelosi, & Fischer, 2013; Van Der Linden, Mathôt,
manipulate tools). Instead, when handling everyday problems (e.g., & Vitu, 2015). In those studies, cues or instructions were given to
hanging a picture on the wall), the supposed reasoning-based agent participants so that their influence on fixation patterns could be in-
might utilise the so-called mechanical knowledge to reason about how ferred. In some studies, black and white images of tools with white
to solve them (i.e., to reason about physical object properties, regard- background were used as visual stimuli, making the observations not
less of whether tools are familiar or not). In so doing, a reasoning-based enough representative of natural gaze behaviour in terms of ecological
agent should produce a mental simulation of tool-use action (ham- validity (e.g., Ambrosini & Costantini, 2017). Furthermore, no eye-
mering a nail in the wall), focusing perception on the best-suited af- tracking studies have considered, so far, the effects, on affordance
fordances for the intended action (Osiurak & Badets, 2016, 2017; perception, deriving from the spatial relationship between objects in
Reynaud, Lesourd, Navarro, & Osiurak, 2016). In a sense, tool use is terms of emerging kinematic properties of the visual scene (e.g., a
here considered as an occurrence of a problem-solving situation sus- hammer-nail pair with the nail far away from the hammer). To our
tained by mechanical reasoning skills (Beck et al., 2011). An attempt to knowledge, only a single eye-tracking study highlighted how the rela-
hypostatise the above argumentations, thereby overcoming the di- tions between the objects may modify the visual encoding of object-tool
chotomy between manipulation-based and reasoning-based ap- pairs (Natraj et al., 2015). Finally, no study has explored whether and,
proaches, has been recently made by Osiurak et al. (2017) who pro- if so, how the visuo-perceptual context may influence the patterns of
posed the Three Action-System model (3AS). On the basis of the dorsal- tool's visual exploration in terms of its functional and mechanical
system partition made by Rizzolatti and Matelli (2003), the three knowledge, nor whether these two distinct kinds of knowledge are re-
neurocognitive systems (3AS) that underlie affordance perception, flected in specific visuo-attentional patterns.
mechanical knowledge and function knowledge are supposed to be, To address the aforementioned issues, we devised an experimental
respectively, the dorso-dorsal system (i.e., the motor control system, in eye-tracking paradigm that allowed us to analyse, for a large time
particular the bilateral superior parietal cortex and the intraparietal window (1750 ms), the gaze behaviour of participants watching dif-
sulcus), the ventro-dorsal system (mainly the left inferior parietal ferent kinds of visual scenes that depicted single tools and object-tool
cortex) and the ventral system (mainly the left temporal cortex; Osiurak pairs. Using the mean fixation duration (expressed in milliseconds) of
et al., 2017; see also: Reynaud et al., 2016; Goldenberg, 2013). different areas of interest (AOIs), we obtained an indirect measure of
According to the direct-visual-route-to-action view, vision guides the amount of visuo-spatial attention allocated by participants to dis-
action (Milner & Goodale, 2008). The allocation of visuo-spatial at- tinct characteristics of the visual scene. These different AOIs are those
tention can be used as an indirect measure of affordance perception. In related to the tool and to the object. In addition, we considered the AOIs
particular, it has been suggested that “implicit recognition of action- related to the “manipulation” and the “functional” parts of the tool, as
related object attributes [affordances] can bias object competition – and they are the parts involved in the use (e.g., the handle of a hammer) and
visual spatial attention” so that “the motor affordance of an object must in the function (e.g., the head of a hammer) of the tool, respectively.
first be recognized, a process that likely involves attention to specific The mean fixation duration of manipulation AOI should be more as-
object features” (Handy, Grafton, Shroff, Ketay, & Gazzaniga, 2003, p. sociated to the action, as manipulation AOI is the part of the tool ac-
424-425). Hence, the correct temporal allocation of visuo-spatial at- tually handled, whereas the mean fixation duration of functional AOI
tention is mandatory to perform visually guided behaviour. Therefore, should be more linked to tool identity and function. This differentiation
if an affordance is perceived and a motor plan is elicited by tool ob- is consistent with that traditionally used in the literature (e.g., Natraj
servation, the fixation pattern of the tool is mostly focused on its action- et al., 2015; Ambrosini & Costantini, 2017; Bub, Masson, & Cree, 2008).
relevant parts (Ambrosini & Costantini, 2017; Handy et al., 2003; Land, To sum up, the primary goal of our investigation was to assess
2006; Natraj, Pella, Borghi, & Wheaton, 2015; Roberts & Humphreys, whether thematic and spatial relations between objects influence the
2011a, 2011b). In that way, affordance perception guides gaze beha- temporal allocation of participants’ visuo-spatial attention by affecting
viour in terms of action, as the representation of the tool use is sup- the action readiness prompted by the visuo-perceptual context. With
posed to be grounded within the motor system (Matheson, White, & action readiness we mean the action possibilities suggested by the
McMullen, 2015; but see also: Martin, 2007). Intriguingly, the motoric context of the visual scene. In particular, the relations between the
representations of tools (grasping, reaching) are strongly activated objects constituting the object-tool pairs were manipulated in such a
when they can be promptly used for functional actions (e.g., Masson, way as to give rise to three conditions. In one condition, the object-tool
Bub, & Breuer, 2011). In other words, tools are represented in terms of pairs were thematically consistent (e.g., hammer-nail) with both objects

3
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

Fig. 1. Example of stimuli. (A) Single-tool condition: a bottle. Tool displayed in the middle of a table. (B) Thematically-consistent condition: a bottle with a glass. A
pair of thematically-consistent objects (tool, the bottle, on the right; object, the glass, on the left). Both are located in the participant’s peri-personal space. (C)
Thematically-inconsistent condition: a bottle and a cap. A pair of thematically-inconsistent objects (tool, the bottle, on the right; object, the cap, on the left). Both are
located in the participant’s peri-personal space. (D) Spatially-inconsistent condition: a bottle and a glass. A pair of thematically consistent but spatially inconsistent
objects. Tool (the bottle) is placed on the right, in the participant’s peri-personal space; object (the glass) is placed on the left, in the participant’s extra-personal
space. In this condition, the object-tool pair is thematically consistent but the represented disposition of the objects in the space does not suggest to use the objects
together, so that we defined this condition as “spatially inconsistent”.

in the peri-personal space of the participant; this condition is supposed visually focusing on the manipulable part of the stimulus, and this
to promote high action readiness. On the other hand, low action should be more evident with a larger time window of analysis
readiness was supposed to be induced in two conditions: thematically- (1750 ms) compared to a shorter one (500 ms).
inconsistent (e.g., hammer-steel pot) pairs, with both objects in the
peri-personal space of the participant, and spatially-inconsistent (e.g.,
2. Methods
hammer-nail) pairs, with the tool (e.g., a hammer) in the peri-personal
space and the object (e.g., a nail) in the extra-personal space of the
2.1. Participants
participant. Additionally, we assessed the temporal dynamics of visual-
exploration patterns for tools, using two different time windows of
Eighteen healthy participants volunteered for participation in the
analysis (500 ms and 1750 ms).
experiment (9 males, mean age = 22.44 years, S.D. = 3.90). All were
Fixation duration, as an indirect measure of visuo-spatial attentional
right handed, based on the Edinburgh Handedness Inventory (Oldfield,
processes, might indeed reflect the degree of information exploration,
1971), had normal or corrected-to-normal vision and self-reported not
such that longer fixation duration should indicate difficulty in ex-
having any prior neurological problems. All were naive as to the pur-
tracting information (Just & Carpenter, 1976). We posited that fixation
poses of the experiment and gave their informed consent. The study
duration on the overall visual scene (tool and object AOIs for the object-
received approval from the Ethics Committee of the Faculty of Educa-
tool pairs) should be affected by the amount of information related to
tion of Suor Orsola Benincasa University.
the action readiness evoked by the visual scene, possibly resulting from
differences in affordance perception. The experimental hypotheses
were therefore based on the three conditions of action readiness of the 2.2. Materials
visual context, with fixation duration expected to be shorter when the
visual context was characterized by higher action readiness than lower Twenty-four three-dimensional (3D) computer-graphics-generated
action readiness. In addition, the action readiness prompted by the stimuli were used in the experiment. In accordance with the experi-
visual scene was predicted to affect the way in which the tool of the mental conditions, four different classes of stimuli were used. The first
object-tool pairs is visually encoded. Specifically, we hypothesised that group of stimuli was composed by 3D color images of a single tool
the tool should be fixated for a longer time on its manipulable part (e.g., placed suspended at the centre of a table with an inclination consistent
the handle of the hammer) when the visual scene includes higher action with a potential use of it by a right-handed person. This group was
readiness, i.e., thematically consistent object-tool pairs (e.g., hammer- composed of six stimuli: a hammer, a bottle opener, a screwdriver, a
nail both within the person’s peri-personal space). Conversely, in a bottle, a whisk, and a spoon. The second group of stimuli was composed
lower action readiness context (thematically/spatially inconsistent ob- by 3D color images containing pairs of objects (a tool on the right – e.g.,
ject-tool pairs), we expected a shorter fixation time on the manipulation a hammer – and an object on the left, e.g., a nail) that were thematically
area of the tool. Finally, following suggestions in the literature consistent, placed on the part of a table closest to the observer, with an
(Ambrosini & Costantini, 2017), we expected that tool’s visual encoding inclination consistent with a potential use of the tool by a right-handed
should start within the first 500 ms with a focus on the functional part person. There were the following six tool-object pairs: hammer-nail,
of the tools. Therefore, we predicted that only when the action re- bottle opener-bottle, screwdriver-screw, bottle-glass, whisk-recipient,
presentation is completed, the fixation pattern will change towards spoon-coffee cup. The third group of stimuli was composed by 3D color
images containing pairs of thematically inconsistent objects (a tool on

4
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

the right – e.g., a whisk – and an object on the left, e.g., a shoe) placed appeared for 3000 ms. After each stimulus, a black screen appeared for
on the part of the table that was closest to the observer, with an in- 4000 ms in order to permit retina relaxation. Each single presentation
clination consistent with a potential use of the tool by a right-handed lasted 7.5 s (500 ms + 3000 ms + 4000 ms). Globally, stimuli pre-
person. This group comprised six tool-object pairs: hammer-steel pot, sentation lasted 180 s (7500 ms × 24 stimuli).
bottle opener-notebook, screwdriver-nut, bottle-cap, whisk-shoe, At the end of the stimulation, a reachability task using the same
spoon-ball. The fourth group of stimuli was composed by 3D color experimental stimuli was administered, which required participants to
images containing pairs of objects (a tool on the right – e.g., a indicate if tools and objects in the visual scene were graspable with the
bottle – and an object on the left, e.g., a glass) that were thematically right hand according to their perspective and to the experimental
consistent but spatially inconsistent. Indeed, the tool was placed on the conditions. In this way, it was possible to ascertain whether the ex-
part of the table that was near the observer, with an inclination con- perimental manipulation of the spatial component (peri-personal and
sistent with a potential use by a right-handed person, while the object extra-personal space, Fig. 1) was successful or not. According to the
was placed on the part of the table that was far from the observer. This experimental conditions described in Fig. 1, in the single tool condition,
group included six tool-object pairs: hammer-nail, bottle opener-bottle, participants correctly reported that the tool was potentially reachable
screwdriver-screw, bottle-glass, whisk-recipient, spoon-coffee cup. and usable using their right hand. For thematically consistent and
Some examples of the stimuli are illustrated in Fig. 1. thematically inconsistent conditions, participants correctly reported
that both the tools and the objects were reachable with their right hand
2.3. Procedure and that the object–tool pair was arranged in the right kinematic way to
allow a potential joint use. Most relevant, in the thematically consistent
The experiment was conducted in the Laboratory of Experimental condition, participants reported that the tool was potentially usable on
Psychology at Suor Orsola Benincasa University (Naples, Italy). The the object (e.g., the hammer was effectively usable on the nail),
room was air-conditioned and the temperature was maintained con- whereas in the thematically inconsistent condition – albeit consistent
stant at 22 °C during the entire duration of the experiment. Light con- from a kinematic point of view – tools and objects were reported to be
ditions of the room were kept stable for all participants and for the not immediately usable together in a proper way (e.g., the bottle opener
entire duration of the experiment. Before starting, the participants was not considered usable on the notebook despite their spatial proxi-
signed informed consent. They were asked to self-report their right- mity). Finally, for the spatially inconsistent condition, participants
handedness, their adequate visual acuity and the absence of any neu- correctly reported that only tools were reachable with their right hand
rological and psychiatric diseases at the date of the experiment. The and, consistently with our experimental manipulation, the object–tool
Edinburgh Handedness Inventory (Oldfield, 1971) was administered to pairs were considered not usable together despite their thematic con-
participants in order to verify that they were actually right-handers. gruency, as the object was spatially located too far from the tool. For
Then, a classic optometric test with participants placed three meters each participant, the overall duration of the experiment was 14 min. At
away from the test stimuli was administered to evaluate visual acuity. the end of the experiment, participants were debriefed regarding the
The participants were seated on a chair and a headrest was used to purposes of the study and the methods used. No participant was ex-
prevent head movements in order to allow a precise eye-tracking re- cluded from the sample.
cording. Participants seated at the distance of 54 cm from the monitor
(27″) and were asked to keep their right hand motionless on the desk. In 2.4. Apparatus
this way, the right hand was resting on the right side of the monitor at
about 15 cm from it, becoming peripherally visible to the participants in In this study, we used a Full-HD Webcam (Logitech HD Pro C920) as
the context of the visual scene. More specifically, the right hand was eye-tracking hardware and the RealEye.io online platform to manage
located at an angle between 35° and 40° of the right visual space (mid- the experiment and to acquire gaze behaviour data. The eye-tracking
peripheral vision). This particular setting was chosen on the basis of a technology used in the present study is based on WebGazer, an eye
pilot study through which it was possible to observe the correct re- tracking JavaScript library (Papoutsaki, Sangkloy, Laskey, Daskalova,
cognition of the objects and their spatial disposition in the peri-personal Huang, & Hays, 2016). All stimuli were presented on a 27″ monitor at a
space vs. extra-personal space using an appropriate and preliminary resolution of 1920*1080px.
reachability task. Then, the experimental instructions were given.
Participants were asked to complete an eye-tracking software calibra- 2.5. Gaze-behaviour data
tion procedure by following with their eyes a white cross that se-
quentially appeared on nine parts of the screen (black background). Gaze-behaviour data were analysed in terms of fixation duration on
Afterwards, participants were asked to “observe what appeared on the the AOIs (in milliseconds). Four different AOIs were defined for the
screen in the most natural way as possible” and the experiment started. analysis of fixation duration: the tool, the object, the manipulation part
A single trial of six images related to each experimental condition was of the tool and the functional part of the tool (Fig. 3).
administered. Thus, twenty-four images were randomly presented ac- The latter two areas were respectively defined as the middle-bottom
cording to the experimental visual flow (Fig. 2): before each stimulus, a part of the tool where to put the hand in order to use it and the middle-
fixation point (white cross over a black background in the bottom area top part of the tool through which it is possible to understand its
of the screen) of 500 ms duration was shown. Then, the stimulus function. Taking into consideration the limits of the eye-tracking

Fig. 2. Experimental flow. Fixation point: 500 ms. Stimulus: 3000 ms. Black screen: 4000 ms.

5
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

Fig. 3. Areas of Interest considered in the study. The AOIs were the following: tool circled in green (T), object circled in violet (O), manipulation area of the tool
circled in red (MT), functional area of the tool circled in blue (FT).

Table 1 For all the analyses, an alpha level of 0.05 were used. Bonferroni
Mean fixation time to tool AOI and object AOI. correction was used for multiple comparisons. All analyses were per-
Fixation Time (ms) – Mean and SD
formed with the open-source statistical software R for Apple Mac OS X
operating system.
Tool AOI Object AOI

Context Thematically consistent 491.42 (139.94) 463.78 (144.92) 2.7. Note

Thematically inconsistent 566.04 (184.48) 599.30 (134.23)
Spatially inconsistent 569.16 (238.91) 694.63 (217.05) A preliminary version of the paradigm used in this article was tested
in the first author’s Master’s thesis, within a joint programme (double
degree) between the University of Campania “Luigi Vanvitelli”
technology used, AOIs were calculated with a perimeter increased by (Caserta, Italy) and the University of Lille 3 (Lille, France).
64 pixels in all directions. All the AOI data were generated using custom
Python scripts. Fixation durations of the AOIs of the stimuli were
3. Results
averaged per condition. According to the results of a pilot study, to
reduce the dispersal effects of data due to participants' visual explora-
3.1. Overall visual exploration
tion, only the first 1750 ms of gaze-behaviour data for each stimulus
were analysed.
Data related to mean fixation time spent to look at the tool and
object AOIs are summarized in Table 1.
2.6. Data analyses
The ANOVA revealed a main effect of the Context on overall fixation
duration, F (2, 34) = 8.24, p < 0.001, ηp2 = 0.33. Post-hoc pair-wise
Four distinct data analyses were performed:
comparisons revealed that fixation duration was higher in both the-

• Overall visual exploration: in order to explore the effects of the con- matically inconsistent (p < 0.05) and spatially inconsistent (p < 0.001)
conditions than in the thematically consistent condition. The effect is
text on the overall exploration of the visual scenario we conducted a
shown in Fig. 4. Conversely, neither main effects of AOIs nor interaction
3 × 2 repeated measure ANOVA with Context as a 3-level factor
were found.
(thematically consistent, thematically inconsistent, spatially incon-
sistent) and AOIs (tool, object) as a 2-level factor, on overall fixation
duration (in milliseconds). 3.2. Tool’s visual exploration
• Tool’s visual exploration: in order to explore the effects of the visuo-
Data related to mean fixation time spent by participants to look at
perceptual context on the way participants looked at the tool, we
conducted a 2 × 3 repeated measure ANOVA with AOIs (manip- the manipulation and functional AOIs of the tool are summarized in
ulation area, functional area) as a 2-level factor and Context (the- Table 2.
matically consistent, thematically inconsistent, spatially incon- The ANOVA revealed a main effect of the AOIs on tool fixation
sistent) as a 3-level factor on tool fixation duration (in milliseconds). duration, F(1, 17) = 16.59, p < 0.001, ηp2 = 0.49. This main effect was
• Single-tool’s visual exploration: two paired-t tests were conducted to due to a longer fixation duration for Functional AOI (M = 300.80 ms,
SD = 142.47) than Manipulation AOI (M = 234.48 ms, SD = 110.63).
explore how participants looked at the tool when it was presented
alone, without objects. The first analysis considered all the time- A significant interaction of Context and AOIs was also found, F (2,
window of gaze-behaviour data (1750 ms), while the second one 34) = 25.51, p < 0.00001, ηp2 = 0.60. Post-hoc pair-wise comparisons
explored only the first 500 ms. revealed that fixation duration for manipulation AOI was longer in the
• First 500 ms of tool’s visual exploration: the same analysis as for thematically consistent condition than thematically inconsistent
(p < 0.05) and spatially inconsistent (p < 0.05) conditions. Instead,
“Tool’s visual exploration” was conducted but restricted to the first
500 ms of visual exploration. functional AOI was fixated longer in both thematically inconsistent
(p < 0.0001) and spatially inconsistent conditions (p < 0.0001) than
thematically consistent condition. In the thematically consistent

6
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

Fig. 4. Overall visual exploration. Object-tool pairs were longer fixated in thematically and spatially inconsistent conditions. Vertical bars denote 0.95 confidence
intervals.

Table 2 4. Discussion
Mean fixation time of manipulation and functional AOIs of the tool.
Fixation Time (ms) – Mean and SD
Our investigation was aimed to assess whether and, if so, how the
degree of action readiness prompted by the visuo-perceptual context
Manipulation AOI Functional AOI could modify the allocation of visuo-spatial attention to different Areas
of Interest (AOIs) of the visual scene. In particular, we created a higher
Context Thematically consistent 299.57 (117.46) 192.57 (66.13)
Thematically inconsistent 202.77 (92.87) 354.09 (108.84)
action-readiness (thematically consistent object-tool pairs) and two
Spatially inconsistent 201.08 (94.87) 355.74 (168.67) lower action-readiness (thematically inconsistent and spatially incon-
sistent object-tool pairs) conditions. In addition, through the analysis of
the mean fixation duration associated with two specific AOIs of the tool
condition, manipulation AOI of tool was fixated longer than functional (the manipulable part and the functional part), we investigated if these
AOI (p < 0.01) whereas in both thematically and spatially inconsistent visuo-attentional differences could be interpreted in terms of functional
conditions functional AOI was fixated longer than manipulation AOI knowledge and mechanical knowledge. Finally, the patterns of visual
(both p < 0.001). The interaction effect is shown in Fig. 5(a). No main exploration of tools were analysed by using the mean fixation duration
effect of Context was found. within different time windows (500 ms and 1750 ms). Results showed
that when the visuo-perceptual context facilitates affordance perception
(i.e., under higher action-readiness conditions), the visual scene was
3.3. Single-tool’s visual exploration encoded in a faster and more suited-for-action way. Interestingly, par-
ticipants’ visual exploration focused on the manipulable part more than
The results of a paired-t test showed that the fixation duration on on the functional part of tools. The second indication from the present
the manipulation AOI of the tool (M = 423.13 ms, SD = 119.38) was study comes from the analyses of the temporal dynamics of the tool’s
significantly longer than on the functional AOI (M = 235.67 ms, visual exploration. When the first 500 ms were taken into account, tools
SD = 104.77; t(17) = 5.70, p < 0.001). The effect is shown in Fig. 5(b). were fixated longer on their functional part, in all experimental con-
When considering only the first 500 ms of stimuli presentation, the ditions. When the time-window of analysis was 1750 ms, tools appeared
results of a second paired-t test indicated that the fixation duration on fixated longer on their manipulable part if they were presented alone
functional AOI of the tool (M = 117.75 ms, SD = 52.60) was sig- and under the thematically consistent condition.
nificantly longer than the manipulation AOI (M = 83.45 ms, Notably, the mean fixation duration of both objects composing the
SD = 39.90; t(17) = 2.12, p < 0.05). The effect is shown in Fig. 5(c). object-tool pairs was higher under thematically and spatially incon-
sistent conditions than under thematically consistent context condi-
tions. The differences in fixation patterns cannot be attributed to dif-
3.4. First 500 ms of tool’s visual exploration ferences in the stimulus salience. The fact that the variance related to
the main effect of the context is not explained by differences in the
Data related to mean fixation time spent by participants to look at mean fixation duration of the tools as compared to the objects suggests
the manipulation and functional AOIs of the tool in the first 500 ms of that the differences in fixation patterns should not be attributed to
stimuli presentation are summarized in Table 3. different stimulus salience as it would be conceivable in the framework
The ANOVA revealed a main effect of the AOIs on tool fixation of biased competition between perceptual objects (Desimone & Duncan,
duration, F(1, 17) = 39.50, p < 0.0001, ηp2 = 0.69. This main effect 1995; Yantis, 2005). In fact, if so, we should have found differences
was due to a longer fixation duration for Functional AOI between the two perceptual stimuli according to their inherent char-
(M = 83.27 ms, SD = 26.34) than Manipulation AOI (M = 62.40 ms, acteristics. These differences were not found even in the condition of
SD = 12.22). The effect is shown in Fig. 5(d). Neither main effect of spatial inconsistency, in which the object of the object-tool pairs, by
Context, nor interaction effects were found. virtue of the three-dimensional graphic representation of its spatial

7
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

Fig. 5. Tool’s visual exploration. (A) In the thematically consistent condition, tools were longer fixated in their manipulation area whereas in the thematically and
spatially inconsistent conditions the tools’ functional area was fixated longer. (B) When individually presented, tools were fixated longer on their manipulation area.
(C) When individually presented and considering only the first 500 ms of stimuli presentation, tools were fixated longer on their functional area. (D) When tools were
part of an object-tool pair and only the first 500 ms of stimuli presentation were considered, they were fixated longer on their functional area. For (A), (B), (C) and
(D) vertical bars denote 0.95 confidence intervals.

Table 3 Mack, 2007). To date, participants temporally allocate the same

Mean fixation time of manipulation and functional AOI of the tool (first amount of visuo-spatial attention to both the tool and the object of the
500 ms). object-tool pairs and this allocation symmetrically increases as the
Fixation Time (ms) – Mean and SD context changes, becoming higher when the context is inconsistent in
terms of what we called action readiness of the visual scene. Critically,
Manipulation AOI Functional AOI the disruption of the action readiness of the scenario (obtained by
manipulating both thematic and spatial relations between the objects
Context Thematically consistent 60.82 (10.08) 82.16 (21.93)
Thematically inconsistent 60.67 (14.16) 88.94 (27.46) composing the object-tool pairs) showed the same trend in terms of
Spatially inconsistent 65.72 (12.08) 78.71 (29.50) overall gaze behaviour. Thus, by excluding the effects of stimulus sal-
ience, we can reasonably conclude that the effect of the context on the
overall visual encoding of the scene is at least in part mediated by af-
arrangement, was considerably smaller compared to the tool. One fordance perception. When the visuo-perceptual context facilitates af-
might argue that the differences in fixation time may depend on the fordance perception, as in the case of thematically consistent object-
different perceptual configuration of the stimuli and their different tool pairs, this is reflected in a faster visual encoding of the overall
salience. However, the lack of variance explanation in terms of differ- scene. Conversely, when affordance perception is negatively impacted
ences between the perceptual stimuli allows us to reasonably exclude a by the lower action readiness of the visuo-perceptual context (thema-
main influence of bottom-up control strategies built around a saliency tically and spatially-inconsistent conditions), a slower visual encoding
map or other low-level factors (Itti & Koch, 2001). Such an inter- of the scene occurred.
pretation is in line with recent findings in the literature related to the From a cognitive perspective, fixation duration, as a direct effect
empirical evaluation that these kinds of models are insufficient to ac- produced by attentional processes, may reflect the degree of explora-
count for human gaze behaviour (Henderson, Brockmole, Castelhano, & tion of the information, so that longer fixation duration might indicate

8
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

difficulty in extracting information (Just & Carpenter, 1976). In this behaviour toward tools (Roberts & Humphreys, 2011a, 2011b). For
sense, the context of a visual scenario might affect affordance percep- instance, an eye-tracking study showed that black and white images
tion and this might imply the engagement of additional or different with a white background of single tools are mainly looked at the
cognitive resources to eventually resolve an informational conflict. This functional area of the tool during the first 500 ms of their presentation.
cognitive involvement – as suggested by the present investigation – may However, this effect is reduced when participants have their hands tied
be signalled by a greater allocation of attentive resources. This per- behind their backs (Ambrosini & Costantini, 2017). Although not eco-
spective is consistent with the behavioural studies that have highlighted logically informative regarding the visual encoding of tools within a
how affordance perception may be modulated by the visuo-perceptual context, the above-mentioned evidence suggests that gaze behaviour is
context (e.g., Borghi et al., 2012), while, at the same time, it emphasises affected by the possibility to act upon an object and, more interestingly
– embracing an ecological point of view – the flexibility and the dy- for our purposes, that a typical fixation pattern starts with the identi-
namism of the high-level cognitive processes involved in the informa- fication of the object through its functional area. In other words, for
tion gathered from the environment. In other words, the present results adult participants, the visual exploration of tools starts with the en-
strongly support the idea that we do not passively learn the relationship coding of their function and this is reflected, coherently with our sug-
between objects, but we generate it in order to “act” in a context gestions, in a visuo-attentional pattern that emphasizes the functional
(Osiurak & Badets, 2016, 2017; Osiurak et al., 2010; Osiurak, 2014). In area of the tool. In some way, it seems as if the functional knowledge of
such a kind of “reasoning-based” cognitive perspective (Osiurak & the tool guided the initial process of affordance perception so that the
Badets, 2016), affordance is seen as one of the information available in first useful information is “what?”, opening the door to mechanical
the environment that can be used by an agent to act. This position is in knowledge whose intrinsic information is “how?”. Such an interpreta-
line with the working memory hypothesis of affordances as regards the tion is consistent with studies with healthy adult participants that have
claim that the nature of the task influences the cognitive workload and, provided evidence of a faster processing and a dominant role of func-
as a consequence, affordance perception (Randerath, Li, Goldenberg, & tion knowledge in object recognition (Garcea & Mahon, 2012; Ni, Liu, &
Hermsdörfer, 2009; Randerath, Goldenberg, Spijkers, Li, & Yu, 2019).
Hermsdörfer, 2011; Randerath, Martin, & Frey, 2013). In accordance with the above arguments, we found that moving the
When the context facilitates affordance perception (i.e., themati- time-window of analysis to the first 500 ms for the single tool condition
cally-consistent condition), visual encoding of the object-tool pairs is changed the fixation pattern toward a predominance of the functional
not only faster but also better (for action purposes). In fact, our results area over the manipulation area. When the tool was individually pre-
showed that the manipulation area of the tools is fixated longer when sented and a longer fixation duration time-window was considered
presented in a thematic and spatially consistent context compared to (1750 ms), the tool was fixated longer on its manipulation part,
when the tools are presented in an inconsistent condition. Conversely, showing a trend similar to that of thematically consistent object-tool
the functional area of the tool obtained more fixations within thematic pairs. When the time-window of analysis was restricted to the first
and spatial inconsistent contexts compared to the consistent context. 500 ms of visual exploration, the fixation pattern was the opposite,
Crucially, in any context conditions, mean fixation time on the ma- privileging the functional part of the tool. Crucially, considering the
nipulation vs. functional areas was significantly different. This is con- first 500 ms of visual exploration, tools were fixated longer on their
sistent with the idea that these areas pertain to different kinds of object functional part in all experimental condition. Thus, it seems reasonable
knowledge (Garcea & Mahon, 2012). We observed a trend inversion in to conclude that, for reasoning-based agents, the initial visual ex-
the way people look at the tool, such that its specific object features ploration of a tool is aimed at gaining its function and identity
(manipulability and functionality) obtained a different and reversed (“what?”) as reflected in a visuo-attentional pattern that emphasizes the
allocation of visuo-spatial attention as the context changed in terms of functional part. Then, in the absence of conflicting context information,
action readiness. Interestingly, we found the same fixation pattern for a mental simulation of the action might be easily made as an effect of
both thematically and spatially-inconsistent conditions. This result the mechanical knowledge (“how?”), hence shifting visual attention
clearly suggests that it is not only the thematic relation between the towards the tool’s manipulation area, in order to use it (i.e., to actualise
objects of the pair that implicitly modulates the visuo-spatial attention the action through the motor system; see Osiurak & Badets, 2016). But
allocated to specific areas of the tool, but also the spatial arrangement what happens when the “what” conflicts with the “how”? What hap-
of the objects. pens when the function knowledge of a hammer seen on a table (it is a
As discussed above, functional knowledge is a kind of semantic hammer, it is generally used with a nail) conflicts with the empirical
knowledge related to a categorical organization of objects according to evidence that a steel pot, and not a nail, is placed next to it? Probably,
different kinds of activities (e.g., if we have to cut an apple, functional assuming the primacy and the “directness” of the action (Xu & Heinke,
knowledge can be helpful “to infer” that we need a knife). Such a kind 2017), when the action readiness is mined by the context, a reasoning-
of knowledge supports what is known as “functional fixedness”, i.e. the based agent might need to change the “how”, in order to accommodate
linking of a tool with a specific function (German & Barrett, 2005; a possible new use (i.e., reasoning about the way to use a hammer with a
Osiurak, 2014; but see also Maier, 1930; Duncker, 1945). Evidence steel pot). Thus, the hammer, with its initially hammering function,
from clinical neuropsychological studies clearly showed that function could be “re-functionalized” into a valid percussion tool to beat on the
knowledge and real tool use can be selectively impaired (e.g., Bartolo, steel pot (see Chrysikou & Thompson-Schill, 2011; Glenberg &
Daumüller, Della Sala, & Goldenberg, 2007), suggesting that the former Robertson, 2000 for creative object uses and novel affordances). Con-
may not be in charge for the understanding of mechanical actions that ceivably, to do that, one needs to stress the mechanical knowledge re-
can be performed when using a tool. Notably, patients with semantic servoir, searching for a new possible use of the tool on the object
dementia, who are unable to access function knowledge, can try to use (Osiurak & Badets, 2016).
different objects, demonstrating a potential, though not proper, utili- It should be remarked that in our experiment we used a task in
zation (e.g., manipulating a hammer from its head) of the tool, sug- which participants were asked to observe the visual scene as naturally
gesting that they can still use mechanical knowledge (Osiurak et al., as possible. Such an ecological task might activate the implicit goal of
2008). A large amount of evidence in the literature corroborated the finding potential mechanical actions between objects and tools. In line
idea that these two types of knowledge rely on different brain areas, with that, participants might firstly have tried to solve a mechanical
with the function knowledge involving the left temporal cortex knowledge issue (“Which kind of mechanical action can I do?”), then to
(Goldenberg, 2013) and the mechanical knowledge involving the left solve a motor control (manipulative) issue (“How should I grasp the
inferior parietal cortex (Goldenberg & Spatt, 2009). It has also been tool to actualize this mechanical action?”). In other words, observing a
suggested that the functional knowledge of the stimulus affects gaze screwdriver-screw pair, a first environment-gathered information might

9
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

be the one linked to the tool’s function knowledge (“What is it? It is a mechanical reasoning skills (Beck et al., 2011). From this point of view,
screwdriver. It is generally used to screw.”). That could explain the neuroimaging evidence of a left inferior prefrontal cortex involvement
higher fixation pattern to the functional area of the tool in the first in tool use, particularly in action planning and execution, appears quite
500 ms. Then, the presence of a screw next to the screwdriver may be interesting (Reynaud et al., 2016). In fact, these cortical areas are lar-
quite informative to resolve in a fast way (i.e., reduced overall fixation gely implicated in high-level executive functions as well as in motor
time) the mechanical knowledge issue (“How can the screwdriver be timing, sequencing and simulation (Bortoletto & Cunnington, 2010;
used with the screw?”). Hence, the motor control issue could be quickly Koechlin & Summerfield, 2007; Stadler et al., 2011). Intriguingly, the
taken into account (“How to grab and manipulate the screwdriver to rostrolateral prefrontal cortex seems to be involved also in such com-
screw?”). This interpretation could clarify the higher fixation pattern to plex human reasoning tasks as relational integration, i.e., considering
the manipulation area of the tool in the thematically consistent con- multiple relations simultaneously (e.g., Christoff et al., 2001). Note that
dition, when a large time-window of analysis is considered. Notice that – assuming the affordance-competition hypothesis as a slightly different
the mechanical knowledge issue may also be solved easily when a single version of the aforementioned “cascade” mechanism for which different
tool is present in the visual scene, in the absence of visuo-perceptual affordances are pre-activated before being selected (Cisek, 2007) – the
conflicting information like a thematically or spatially inconsistent frontal-lobes involvement might also signal processes that, from the
object. That could explain the same visuo-attentional pattern found for multiple environment-available affordances, select only those that are
both thematically consistent and single-object conditions. On the con- relevant to the individual's intentions, i.e., an inhibitory mechanisms
trary, when a nut and not a screw appears next to a screwdriver, the (see also: Osiurak & Badets, 2016). However, in the absence of a clear
mechanical knowledge issue may not be so easily solved. Then, an neuropsychological correlation between frontal lesions and tool-use
observer could spend considerable time to extract a new possible me- impairments, future research should detail the involvement of the
chanical action between objects of the pair (i.e., increasing fixation frontal cortex in human tool use (Reynaud et al., 2016).
time), while not spending time to solve the motor control issue, with the To sum up, our investigation highlighted, for the first time, differ-
consequent lower fixations on the manipulable part of the tool (i.e., ences in visual-encoding modalities of tools and object-tool pairs as the
visual attention remains fixed to the functional part of the tool). degree of action readiness evoked by the visuo-perceptual context
Such a kind of function-to-mechanical-to-motor “cascade” cognitive changes. Tools were faster and better (for action purposes) visually
mechanism appears to be supported by most recent research on rea- encoded when the visual scene prompted higher action-readiness, i.e.,
soning-based tool use and, particularly, by the recently proposed Three in the thematically consistent condition. In addition, consistently with
Action-System model (Osiurak et al., 2017; Osiurak & Badets, 2016, recent evidence (Ambrosini & Costantini, 2017), we found that tool’s
2017; Reynaud et al., 2016). In this article, we suggest that the peculiar visual exploration starts with focusing on the tool’s functional area,
visuo-attentional patterns highlighted by our investigation might reflect even when the tool is part of an object-tool pair. Then, when the visual
the interactions between functional knowledge, mechanical knowledge scene is characterized by higher action readiness (e.g., hammer and
and motor control neurocognitive systems. More precisely, mechanical nail) or when no conflicting information has to be processed (e.g.,
knowledge (i.e., the ventro-dorsal system) might be seen as a bridge single tool) the manipulation area of the tool is looked at significantly
between high-level semantic information pertaining object identity and longer. As discussed above, by showing implicit, peculiar and flexible
functions (i.e., the ventral system) and the motor control system (i.e., modifications of the context-consistent visuo-attentional patterns, our
dorso-dorsal system), thus generating a mental simulation of the action results appear to converge towards recent research lines that propose a
related to tool use and guiding the perception of the corresponding reasoning-based approach related to human tool use (Osiurak & Badets,
affordances. This is consistent with recent evidence indicating, as 2016, 2017). Such a kind of visuo-attentional patterns seem to reflect
neural correlates of tool use, a wide and complex interplay of brain the engagement of different and complex neurocognitive systems in-
regions pertaining to the fronto-parietal and occipito-temporal net- volved in human tool-use processing (De Bellis et al., 2018; Lambon
works (Reynaud et al., 2016; but see also: Beauchamp & Martin, 2007; Ralph et al., 2017; Reynaud et al., 2016). In a theoretical perspective
Boronat et al., 2005; Canessa et al., 2007; Johnson-Frey, 2004; Chao & that assumes the centrality of action and within a cognitive and rea-
Martin, 2000; Creem-Regehr & Lee, 2005). Specifically, mechanical soning-based perimeter, we propose to call “action reappraisal” the
knowledge appears to be stored in the left inferior parietal cortex, process generated (and presumably reflected by the visuo-attentional
particularly within the cytoarchitectonic area PF of the supramarginal patterns we found) by the set of neurocognitive systems that – in a
gyrus (SMG), whereas the neurocognitive system underlying affordance dynamic way – utilises multiple sources of information (e.g., affor-
perception (i.e., the motor control system) involves the bilateral su- dances, object knowledge, mechanical knowledge, functional knowl-
perior parietal cortex and the intraparietal sulcus (putative human edge, context conditions, abstract knowledge, etc.) to reason about the
anterior intraparietal sulcus area and the anterior dorsal IPS). An in- “possibility to act” within and on a context, in a proper and agent’s
tegrative neurocognitive layer between mechanical knowledge and the consistent-with-intentions way. In other words, we support the idea
motor control system seems to be the left anterior portion of SMG, that a reasoning-based agent uses tools to solve everyday problems,
extending to the cytoarchitectonic area PFt of SMG (Caspers et al., thus acquiring, updating and reasoning about the most appropriate use
2006, 2008; Reynaud et al., 2016). Neural counterparts of functional of tools in a context (Osiurak & Badets, 2016, 2017), rather than to
knowledge appear to be rather debated in the literature (e.g., Boronat passively learn and actualise the actions that can be performed with
et al., 2005; Canessa et al., 2007; Goldenberg, 2013). However, recent them. Although convergent and recent evidence in the field of the
evidence has suggested that the left temporal cortex, the left posterior reasoning-based approach to human tool use seems to support the ac-
middle temporal gyrus (pMTG) and the lateral occipital complex (LOC), tion reappraisal idea (e.g., Osiurak et al., 2017; Osiurak & Badets, 2016,
may be assumed as neural substrates of functional knowledge 2017; Reynaud et al., 2016; De Bellis et al., 2018; Lambon Ralph et al.,
(Goldenberg, 2013; Orban & Caruana, 2014; Reynaud et al., 2016; 2017), and despite its appeal, further studies are clearly necessary in
Roux-Sibilon et al., 2018). Crucially, increasing evidence in the litera- order to confirm its validity.
ture indicates the inferior parietal cortex and the middle temporal areas
as regions where a multimodal integration of action and semantic in-
formation takes place to generate high-level cognitive representations Declaration of Competing Interest
about tools (De Bellis et al., 2018; Reynaud et al., 2016 but see also:
Lambon Ralph, Jefferies, Patterson, & Rogers, 2017). Finally, in a the- None.
oretical perspective that emphasizes a reasoning-based approach, we
have considered tool use as a kind of problem-solving task sustained by

10
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

References Goldenberg, G. (2013). Apraxia: The cognitive side of motor control. Oxford: Oxford
University Press.
Goldenberg, G., & Randerath, J. (2015). Shared neural substrates of apraxia and aphasia.
Ambrosini, E., & Costantini, M. (2017). Body posture differentially impacts on visual Neuropsychologia, 75, 40–49.
attention towards tool, graspable, and non-graspable objects. Journal of Experimental Goldenberg, G., & Spatt, J. (2009). The neural basis of tool use. Brain, 132(6), 1645–1655.
Psychology: Human Perception and Performance, 43(2), 360. Green, C., & Hummel, J. E. (2006). Familiar interacting object pairs are perceptually
Bartolo, A., Daumüller, M., Della Sala, S., & Goldenberg, G. (2007). Relationship between grouped. Journal of Experimental Psychology: Human Perception and Performance,
object-related gestures and the fractionated object knowledge system. Behavioural 32(5), 1107.
Neurology, 18(3), 143–147. Grezes, J., & Decety, J. (2002). Does visual perception of object afford action? Evidence
Beauchamp, M. S., & Martin, A. (2007). Grounding object concepts in perception and from a neuroimaging study. Neuropsychologia, 40(2), 212–222.
action: Evidence from fMRI studies of tools. Cortex, 43(3), 461–468. Handy, T. C., Grafton, S. T., Shroff, N. M., Ketay, S., & Gazzaniga, M. S. (2003). Graspable
Beck, S. R., Apperly, I. A., Chappell, J., Guthrie, C., & Cutting, N. (2011). Making tools objects grab attention when the potential for action is recognized. Nature
isn’t child’s play. Cognition, 119(2), 301–306. Neuroscience, 6(4), 421.
Borghi, A. M., Flumini, A., Natraj, N., & Wheaton, L. A. (2012). One hand, two objects: Henderson, J. M., Brockmole, J. R., Castelhano, M. S., & Mack, M. (2007). Visual saliency
Emergence of affordance in contexts. Brain and Cognition, 80(1), 64–73. does not account for eye movements during visual search in real-world scenes. In R.
Boronat, C. B., Buxbaum, L. J., Coslett, H. B., Tang, K., Saffran, E. M., Kimberg, D. Y., & P. G. Van Gomple, M. H. Fischer, W. S. Murray, & R. L. Hill (Eds.). Eye Movements: A
Detre, J. A. (2005). Distinctions between manipulation and function knowledge of window on mind and brain (pp. 537–562). Oxford, United Kingdom: Elsevier.
objects: Evidence from functional magnetic resonance imaging. Cognitive Brain Humphreys, G. W., Yoon, E. Y., Kumar, S., Lestou, V., Kitadono, K., Roberts, K. L., &
Research, 23(2–3), 361–373. Riddoch, M. J. (2010). The interaction of attention and action: From seeing action to
Bortoletto, M., & Cunnington, R. (2010). Motor timing and motor sequencing contribute acting on perception. British Journal of Psychology, 101(2), 185–206.
differently to the preparation for voluntary movement. Neuroimage, 49(4), Itti, L., & Koch, C. (2001). Computational modelling of visual attention. Nature Reviews
3338–3348. Neuroscience, 2(3), 194.
Bub, D. N., Masson, M. E., & Cree, G. S. (2008). Evocation of functional and volumetric Jeannerod, M., Arbib, M. A., Rizzolatti, G., & Sakata, H. (1995). Grasping objects: The
gestural knowledge by objects and words. Cognition, 106, 27–58. cortical mechanisms of visuomotor transformation. Trends in Neurosciences, 18(7),
Buxbaum, L. J. (2001). Ideomotor apraxia: A call to action. Neurocase, 7(6), 445–458. 314–320.
Buxbaum, L. J., & Kalénine, S. (2010). Action knowledge, visuomotor activation, and Johnson-Frey, S. H. (2004). The neural bases of complex tool use in humans. Trends in
embodiment in the two action systems. Annals of the New York Academy of Sciences, Cognitive Sciences, 8(2), 71–78.
1191(1), 201–218. Just, M. A., & Carpenter, P. A. (1976). Eye fixations and cognitive processes. Cognitive
Buxbaum, L. J., & Saffran, E. M. (2002). Knowledge of object manipulation and object Psychology, 8(4), 441–480.
function: Dissociations in apraxic and nonapraxic subjects. Brain and Language, 82(2), Kalénine, S., Mirman, D., Middleton, E. L., & Buxbaum, L. J. (2012). Temporal dynamics
179–199. of activation of thematic and functional knowledge during conceptual processing of
Canessa, N., Borgo, F., Cappa, S. F., Perani, D., Falini, A., Buccino, G., ... Shallice, T. manipulable artifacts. Journal of Experimental Psychology: Learning, Memory, and
(2007). The different neural correlates of action and functional knowledge in se- Cognition, 38(5), 1274.
mantic memory: An FMRI study. Cerebral Cortex, 18(4), 740–751. Kalénine, S., Peyrin, C., Pichat, C., Segebarth, C., Bonthoux, F., & Baciu, M. (2009). The
Caspers, S., Eickhoff, S. B., Geyer, S., Scheperjans, F., Molhberg, H., Zilles, K., & Amunts, sensory-motor specificity of taxonomic and thematic conceptual relations: A beha-
K. (2008). The human inferior parietal lobule in stereotaxic space. Brain Structure & vioral and fMRI study. Neuroimage, 44(3), 1152–1162.
Function, 212, 481–495. Koechlin, E., & Summerfield, C. (2007). An information theoretical approach to prefrontal
Caspers, S., Geyer, S., Schleicher, A., Mohlberg, H., Amunts, K., & Zilles, K. (2006). The executive function. Trends in Cognitive Sciences, 11(6), 229–235.
human inferior parietal cortex: Cytoarchitectonic parcellation and interindividual Króliczak, G., & Frey, S. H. (2009). A common network in the left cerebral hemisphere
variability. Neuroimage, 33, 430–448. represents planning of tool use pantomimes and familiar intransitive gestures at the
Chao, L. L., & Martin, A. (2000). Representation of manipulable man-made objects in the hand-independent level. Cerebral Cortex, 19(10), 2396–2410.
dorsal stream. Neuroimage, 12(4), 478–484. Labeye, E., Oker, A., Badard, G., & Versace, R. (2008). Activation and integration of motor
Christoff, K., Prabhakaran, V., Dorfman, J., Zhao, Z., Kroger, J. K., Holyoak, K. J., & components in a short-term priming paradigm. Acta Psychologica, 129(1), 108–111.
Gabrieli, J. D. (2001). Rostrolateral prefrontal cortex involvement in relational in- Lambon Ralph, M. A. L., Jefferies, E., Patterson, K., & Rogers, T. T. (2017). The neural and
tegration during reasoning. Neuroimage, 14(5), 1136–1149. computational bases of semantic cognition. Nature Reviews Neuroscience, 18(1), 42.
Chrysikou, E. G., & Thompson-Schill, S. L. (2011). Dissociable brain states linked to Land, M. F. (2006). Eye movements and the control of actions in everyday life. Progress in
common and creative object use. Human Brain Mapping, 32(4), 665–675. Retinal and Eye Research, 25(3), 296–324.
Cisek, P. (2007). Cortical mechanisms of action selection: the affordance competition Maier, N. R. (1930). Reasoning in humans. I. On direction. Journal of Comparative
hypothesis. Philosophical Transactions of the Royal Society B: Biological Sciences, Psychology, 10(2), 115.
362(1485), 1585–1599. Martin, A. (2007). The representation of object concepts in the brain. Annual Review of
Collette, C., Bonnotte, I., Jacquemont, C., Kalénine, S., & Bartolo, A. (2016). The devel- Psychology, 58, 25–45.
opment of object function and manipulation knowledge: Evidence from a semantic Masson, M. E., Bub, D. N., & Breuer, A. T. (2011). Priming of reach and grasp actions by
priming study. Frontiers in Psychology, 7, 1239. handled objects. Journal of Experimental Psychology: Human Perception and
Costantini, M., Ambrosini, E., Tieri, G., Sinigaglia, C., & Committeri, G. (2010). Where Performance, 37, 1470–1484.
does an object trigger an action? An investigation about affordances in space. Matheson, H., White, N., & McMullen, P. (2015). Accessing embodied object re-
Experimental Brain Research, 207(1–2), 95–103. presentations from vision: A review. Psychological Bulletin, 141(3), 511.
Creem-Regehr, S. H., & Lee, J. N. (2005). Neural representations of graspable objects: Are McNair, N. A., & Harris, I. M. (2014). The contextual action relationship between a tool
tools special? Cognitive Brain Research, 22(3), 457–469. and its action recipient modulates their joint perception. Attention, Perception, &
De Bellis, F., Magliacano, A., Sagliano, L., Conson, M., Grossi, D., & Trojano, L. (2018). Psychophysics, 76(1), 214–229.
Left inferior parietal and posterior temporal cortices mediate the effect of action Milner, A. D., & Goodale, M. A. (2008). Two visual systems re-viewed. Neuropsychologia,
observation on semantic processing of objects: Evidence from rTMS. Psychological 46(3), 774–785.
Research Psychologische Forschung, 1–14. Mizelle, J. C., & Wheaton, L. A. (2010). Why is that hammer in my coffee? A multimodal
De Stefani, E., Innocenti, A., Bernardi, N. F., Campione, G. C., & Gentilucci, M. (2012). imaging investigation of contextually based tool understanding. Frontiers in Human
The bottle and the glass say to me:“Pour!” Experimental Brain Research, 218(4), Neuroscience, 4, 233.
539–549. Mounoud, P. (1996). A recursive transformation of central cognitive mech- anisms: The
Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective visual attention. shift from partial to whole representation. In A. J. Sameroff, & M. M. Haith (Eds.). The
Annual Review of Neuroscience, 18(1), 193–222. five to seven year shift: The age of reason and responsibility (pp. 85–110). Chicago, IL:
Duncker, K. (1945). On problem-solving (L. S. Lees, Trans.). Psychological Monographs, Chicago University Press.
58(5) i-113. Mounoud, P., Duscherer, K., Moy, G., & Perraudin, S. (2007). The influence of action
Ellis, R., & Tucker, M. (2000). Micro-affordance: The potentiation of components of action perception on object recognition: A developmental study. Developmental Science,
by seen objects. British Journal of Psychology, 91(4), 451–471. 10(6), 836–852.
Estes, Z., Golonka, S., & Jones, L. L. (2011). Thematic thinking: The apprehension and Myachykov, A., Ellis, R., Cangelosi, A., & Fischer, M. H. (2013). Visual and linguistic cues
consequences of thematic relations. Psychology of Learning and Motivation – Advances to graspable objects. Experimental Brain Research, 229(4), 545–559.
in Research and Theory, 54, 249. Myung, J. Y., Blumstein, S. E., & Sedivy, J. C. (2006). Playing on the typewriter, typing on
Garcea, F. E., & Mahon, B. Z. (2012). What is in a tool concept? Dissociating manipulation the piano: Manipulation knowledge of objects. Cognition, 98(3), 223–243.
knowledge from function knowledge. Memory & Cognition, 40(8), 1303–1313. Natraj, N., Pella, Y. M., Borghi, A. M., & Wheaton, L. A. (2015). The visual encoding of
German, T. P., & Barrett, H. C. (2005). Functional fixedness in a technologically sparse tool–object affordances. Neuroscience, 310, 512–527.
culture. Psychological Science, 16(1), 1–5. Ni, L., Liu, Y., & Yu, W. (2019). The dominant role of functional action representation in
Gibson, J. J. (1977). The theory of affordance. In R. E. Shaw, & J. Bransford (Eds.). object recognition. Experimental Brain Research, 237(2), 363–375.
Perceiving, acting, and knowing: Toward an ecological psychology (pp. 62–82). Hillsdale, Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh in-
NJ: Erlbaum. ventory. Neuropsychologia, 9(1), 97–113.
Gibson, J. J. (1979). The ecological approach to visual perception. Boston, MA: Houghton Orban, G. A., & Caruana, F. (2014). The neural basis of human tool use. Frontiers in
Mifflin Co. Psychology, 5, 310.
Glenberg, A. M., & Robertson, D. A. (2000). Symbol grounding and meaning: A com- Osiurak, F. (2014). What neuropsychology tells us about human tool use? The four
parison of high-dimensional and embodied theories of meaning. Journal of Memory constraints theory (4CT): Mechanics, space, time, and effort. Neuropsychology Review,
and Language, 43(3), 379–401. 24(2), 88–115.

11
G. Federico and M.A. Brandimonte Brain and Cognition 135 (2019) 103582

Osiurak, F., & Badets, A. (2016). Tool use and affordance: Manipulation-based versus motor cortex. Human Brain Mapping, 30(4), 1168–1187.
reasoning-based approaches. Psychological Review, 123(5), 534. Rizzolatti, G., & Matelli, M. (2003). Two different streams form the dorsal visual system:
Osiurak, F., & Badets, A. (2017). Use of tools and misuse of embodied cognition: Reply to Anatomy and functions. Experimental Brain Research, 153(2), 146–157.
Buxbaum (2017). Psychological Review, 124(3), 361–368. Roberts, K. L., & Humphreys, G. W. (2011a). Action relations facilitate the identification
Osiurak, F., Aubin, G., Allain, P., Jarry, C., Richard, I., & Le Gall, D. (2008). Object uti- of briefly-presented objects. Attention, Perception, & Psychophysics, 73(2), 597–612.
lization and object usage: A single-case study. Neurocase, 14(2), 169–183. Roberts, K. L., & Humphreys, G. W. (2011b). Action-related objects influence the dis-
Osiurak, F., Jarry, C., & Le Gall, D. (2010). Grasping the affordances, understanding the tribution of visuospatial attention. The Quarterly Journal of Experimental Psychology,
reasoning: Toward a dialectical theory of human tool use. Psychological Review, 64(4), 669–688.
117(2), 517. Roux-Sibilon, A., Kalénine, S., Pichat, C., & Peyrin, C. (2018). Dorsal and ventral stream
Osiurak, F., Rossetti, Y., & Badets, A. (2017). What is an affordance? 40 years later. contribution to the paired-object affordance effect. Neuropsychologia, 112, 125–134.
Neuroscience & Biobehavioral Reviews, 77, 403–417. Schreuder, R., d'Arcais, G. B. F., & Glazenborg, G. (1984). Effects of perceptual and
Papoutsaki, A., Sangkloy, P., Laskey, J., Daskalova, N., Huang, J., & Hays, J. (2016). conceptual similarity in semantic priming. Psychological Research Psychologische
Webgazer: Scalable webcam eye tracking using user interactions. In Proceedings of Forschung, 45(4), 339–354.
the twenty-fifth international joint conference on artificial intelligence-IJCAI 2016. Stadler, W., Schubotz, R. I., Von Cramon, D. Y., Springer, A., Graf, M., & Prinz, W. (2011).
Perraudin, S., & Mounoud, P. (2009). Contribution of the priming paradigm to the un- Predicting and memorizing observed action: Differential premotor cortex involve-
derstanding of the conceptual developmental shift from 5 to 9 years of age. ment. Human Brain Mapping, 32(5), 677–687.
Developmental Science, 12(6), 956–977. Thill, S., Caligiore, D., Borghi, A. M., Ziemke, T., & Baldassarre, G. (2013). Theories and
Pfurtscheller, G., & Da Silva, F. L. (1999). Event-related EEG/MEG synchronization and computational models of affordance and mirror systems: An integrative review.
desynchronization: Basic principles. Clinical Neurophysiology, 110(11), 1842–1857. Neuroscience & Biobehavioral Reviews, 37(3), 491–521.
Pluciennicka, E., Wamain, Y., Coello, Y., & Kalénine, S. (2016). Impact of action primes Tucker, M., & Ellis, R. (2001). The potentiation of grasp types during visual object ca-
on implicit processing of thematic and functional similarity relations: Evidence from tegorization. Visual Cognition, 8(6), 769–800.
eye-tracking. Psychological Research Psychologische Forschung, 80(4), 566–580. Van Der Linden, L., Mathôt, S., & Vitu, F. (2015). The role of object affordances and center
Randerath, J., Goldenberg, G., Spijkers, W., Li, Y., & Hermsdörfer, J. (2011). From pan- of gravity in eye movements toward isolated daily-life objects. Journal of Vision,
tomime to actual use: How affordances can facilitate actual tool-use. 15(5), 8 8-8.
Neuropsychologia, 49(9), 2410–2416. Wamain, Y., Gabrielli, F., & Coello, Y. (2016). EEG μ rhythm in virtual reality reveals that
Randerath, J., Li, Y., Goldenberg, G., & Hermsdörfer, J. (2009). Grasping tools: Effects of motor coding of visual objects in peripersonal space is task dependent. Cortex, 74,
task and apraxia. Neuropsychologia, 47(2), 497–505. 20–30.
Randerath, J., Martin, K. R., & Frey, S. H. (2013). Are tool properties always processed Wamain, Y., Pluciennicka, E., & Kalénine, S. (2015). A saw is first identified as an object
automatically? The role of tool use context and task complexity. Cortex, 49(6), used on wood: ERP evidence for temporal differences between Thematic and
1679–1693. Functional similarity relations. Neuropsychologia, 71, 28–37.
Reynaud, E., Lesourd, M., Navarro, J., & Osiurak, F. (2016). On the neurocognitive origins Xu, S., & Heinke, D. (2017). Implied between- object actions affect response selection
of human tool use: A critical review of neuroimaging data. Neuroscience & without knowledge about object functionality. Visual Cognition, 1–17.
Biobehavioral Reviews, 64, 421–437. Yantis, S. (2005). How visual salience wins the battle for awareness. Nature Neuroscience,
Riddoch, M. J., Humphreys, G. W., Heslop, J., & Castermans, E. (2002). Dissociations 8(8), 975.
between object knowledge and everyday action. Neurocase, 8(1), 100–110. Yee, E., Huffstetler, S., & Thompson-Schill, S. L. (2011). Function follows form: Activation
Riddoch, M. J., Humphreys, G. W., Hickman, M., Clift, J., Daly, A., & Colin, J. (2006). I of shape and function features during object identification. Journal of Experimental
can see what you are doing: Action familiarity and affordance promote recovery from Psychology: General, 140(3), 348.
extinction. Cognitive Neuropsychology, 23(4), 583–605. Yoon, E. Y., Humphreys, G. W., & Riddoch, M. J. (2010). The paired-object affordance
Ritter, P., Moosmann, M., & Villringer, A. (2009). Rolandic alpha and beta EEG rhythms' effect. Journal of Experimental Psychology: Human Perception and Performance, 36(4),
strengths are inversely related to fMRI-BOLD signal in primary somatosensory and 812.

12