ArticlePDF Available

The artist emerges: Visual art learning alters neural structure and function

November 2014
NeuroImage 105

DOI:10.1016/j.neuroimage.2014.11.014

Source
PubMed

Authors:

Alexander A Schlegel

Sergey V Fogelson

Dartmouth College

Xueting Li

Renmin University of China

Show all 9 authorsHide

Experimental design. A. Areas of cognition important to the production of represen-

…

Figures - uploaded by Alexander A Schlegel

Content may be subject to copyright.

Content uploaded by Alexander A Schlegel

Content may be subject to copyright.

The artist emerges: Visual art learning alters neural structure

and function

Alexander Schlegel

⁎, Prescott Alexander

, Sergey V. Fogelson

,XuetingLi

a,b

,ZhengangLu

, Peter J. Kohler

Enrico Riley

, Peter U. Tse

,MingMeng

Dartmouth College, Department of Psychological and Brain Sciences, Hanover, NH 03755, USA

Beijing Normal University, National Key Laboratory of Cognitive Neuroscience and Learning, Beijing 100875, China

Dartmouth College, Studio Art Department, Hanover, NH 03755, USA

abstractarticle info

Article history:

Accepted 7 November 2014

Available online 15 November 2014

Keywords:

Art

Creative cognition

Perception

Perception-to-action

Plasticity

White matter

Cerebellum

Motor cortex

DTI

fMRI

MVPA

How does the brain mediate visual artistic creativity? Here we studied behavioraland neural changes indrawing

and painting students compared to students who did not study art. We investigated three aspects of cognition

vital to many visual artists: creative cognition, perception, and perception-to-action. We found that the art stu-

dents became more creative via the reorganization of prefrontal white matter but did not ﬁnd any signiﬁcant

changes in perceptual ability or related neural activity in the art studentsrelative to the control group. Moreover,

the art students improved in their ability to sketch human ﬁgures from observation, and multivariate patterns of

cortical and cerebellar activity evoked by this drawing task became increasingly separable between art and non-

art students. Our ﬁndings suggest that the emergence of visual artistic skills is supported by plasticity in neural

pathways that enable creative cognition and mediate perceptuomotor integration.

Introduction

Art is a complex and uniquely human phenomenon. The creation of

artistic work has historically been a mysterious and poorly understood

process, often even by artists themselves (Stiles and Selz, 2012). How-

ever, according to central tenets of neuroscience, the work of an artist

must be mediated by the brain. How does the brain support the cogni-

tive skills necessary to create art?

Art has appeared in many forms throughout human history. The qual-

ities that distinguish artistic work are thus often difﬁcult to deﬁne. For ex-

ample, while some trompe l'oeil painters such as William Harnett

(Frankenstein, 1953) attain an astounding ability to recreate visual scenes

accurately, representation for other painters such as abstract expression-

ist Barnett Newman (Shiff, 2004) is less important than the concepts or

processes that their works communicate. Nonetheless, most artists, re-

gardless of their motivation or medium, spend years developing patterns

of thought and behavior that lead ultimately to expression in a work of

art. Here, we focused narrowly on a single type of artwork: representa-

tional, two-dimensional visual depictions created from observation. We

necessarily ignored many important factors such as social, cultural, and

affective contexts that are vital to the work of many artists (cf. Stiles

and Selz, 2012). No single study can address every factor that inﬂuences

artistic skill. However, the results presented here may provide a window

into some of the neural processes that endow humans with a seemingly

limitless ability to create new objects, ideas, and processes.

In the current study we investigated how artistic behaviors are

learned, focusing on representational visual art and on three areas of cog-

nition that are relevant to many visual artists: creative cognition, visual

perception, and perception-to-action (Fig. 1A). We asked how skills asso-

ciated with each of these three cognitive domains change and how the

brain reorganizes as students learn to create visual art. We recruited 35

undergraduate college students for monthly testing; 17 of these partici-

pants took a 3-month-long introductory observational drawing or paint-

ing course offered by the Studio Art Department at Dartmouth College,

while 18 control participants did not study art. All participants attended

monthly MRI scanning sessions. Below we introduce the three areas of

cognition that we studied and their potential relevance to visual art.

Creative cognition

Artists are distinguished by the ability to think in new ways, devel-

oping new patterns of and connections between ideas to imagine and

NeuroImage 105 (2015) 440–451

⁎Corresponding author at: Department of Psychological and Brain Sciences, H. B. 6207,

Moore Hall, Dartmouth College, Hanover, NH 03755, USA. Fax: +1 603 646 1419.

E-mail address: schlegel@gmail.com (A. Schlegel).

http://dx.doi.org/10.1016/j.neuroimage.2014.11.014

Contents lists available at ScienceDirect

NeuroImage

journal homepage: www.elsevier.com/locate/ynimg

create artifacts and processes that have never existed previously. The

sources of this creativity are among the least understood and most my-

thologized aspects of art production (Milbrandt and Milbrandt, 2011;

Taylor, 1976). Creative cognition is notoriously difﬁcult to deﬁne within

ascientiﬁc context, partly because creativity can be manifested in myr-

iad domains such as artistic and scientiﬁcﬁelds, verbal and visual mo-

dalities, and divergent and convergent thought (Dietrich and Kanso,

2010). Many questions about what makes artists creative remain

open, especially with respect to the brain's role in these creative pro-

cesses. Previous neuroscientiﬁc studies have used a range of approaches

to study the neural basis of creativity in artists, but little consensus

about this basis has emerged. For example, Bhattacharya and Petsche

(2005) used EEG to study differences in cortical activity between artists

and non-artists as both produced drawings of their own choice and

found differences in short and long range neural synchronization pat-

terns between the two groups. Kowatari et al. (2009) asked design ex-

perts and novice participants to invent a new type of pen while

undergoing functional MRI (fMRI) scans and found that creative output

was correlated with the degree of dominance of right over left prefron-

tal cortical activity. Limb and Braun (2008) used fMRI to show that jazz

pianists experienced extensive deactivation of prefrontal cortex when

they played improvised compared to over-learned musical pieces.

Solso (2001), on the other hand, found reduced activity in the parietal

cortex in a skilled portrait artist compared to a novice participant as

both produced drawings of faces. While little consensus has emerged

from such studies, the many emerging ﬁndings about both artists and

creative cognition more generally have shown that creativity is a com-

plex rather than monolithic process and that researchers must therefore

avoid the tendency to reduce creativity to simple conceptual constructs

(Arden et al., 2010; Dietrich and Kanso, 2010; Hee Kim, 2006). Thus, in

this study we chose assessments of creativity (described below) that

measured many aspects of creative ability.

A recent study by Jung et al. (2010) investigated the relationship be-

tween white matter organization and creative cognitive ability using

diffusion tensor imaging (DTI). They found an inverse correlation in

the frontal lobes between fractional anisotropy (FA; a measure of the di-

rectionality of water diffusion in white matter) and both divergent

thinkingand openness, such that more creative individuals as measured

by these traits tended to have lower frontal white matter FA. While FAis

often associated with myelination of axons, several other properties

such as axon count, axon packing, and crossing ﬁbers can affect the an-

isotropy of water diffusion in the brain as well. While the exact neural

correlates of FA are not determined precisely, DTI nonetheless provides

a non-invasive means of investigating longitudinal changes in the struc-

ture of white matter. Several recent studies have shown that diffusion

tensor imaging (DTI) is an effective tool for tracking learning-related

changes in the white matter organization of the brain in as little as six

weeks or as long as nine months (May, 2011; Schlegel et al., 2012;

Scholz et al., 2009). Even shorter term changes (in as little as ﬁve

days) have been observed in brainstructure using other imaging modal-

ities (Ditye et al., 2013; Driemeyer et al., 2008).

Visual perception

The visual system is organized to recover intrinsic properties of per-

ceived objects such as size andreﬂectance. It therefore often counteracts

context-dependent aspects of objects such as distance from the observ-

er and ambient luminance (Todorović, 2002). In other words, the brain

constructs our perception of the world not necessarily in accordance

with the physical stimulation, but rather as it infers things to be intrin-

sically.For instance, a white ﬂower still appears white in a blue-lit room,

even though the ﬂower reﬂects only blue light in such a room. While

such inferences on the part of the visual system permit us to perceive in-

trinsic properties of objects (e.g. size, shape, or pigment), they can also

lead to illusory percepts such as the Craik-O'Brien-Cornsweet and Müll-

er-Lyer illusions (Figs. 1C & D) (Müller-Lyer, 1889; Todorović,1987).

Fig. 1. Experimental design. A. Areas of cognition important to the production of represen-

tational visual art: creative cognition, perception, and perception-to-action pathways. B.

Example prompt and participant response from the TTCT. Participants were instructed

to draw something that no one else would think of, to tell as rich and complete a story

as possible, and to give each response a creative title. C. The Craik–O'Brien–Cornsweet il-

lusion,in which a dark-to-light gradientheading left of centerand a light-to-dark gradient

heading right of center cause a uniform gray rectangle to appear darker on the left side

than on the right side. D. The Müller–Lyer illusion, in which a line segment with outward

facing arrow ends appearslonger than an identical-length line segment with inward fac-

ing arrow ends. E. Example stimulus and participant response from the gesture drawing

task. Participants had 30 s to complete a gesture drawing of the observed human ﬁgures.

441A. Schlegel et al. / NeuroImage 105 (2015) 440–451

One skill acquired by many visual artists is the ability to create pre-

cise, realistic representations of the world. For these representations to

appear realistic, they must reﬂect accurately the physical, rather than

the inferred, properties of the observed environment. A representation-

al artist may therefore need to counteract these inferences. Otherwise,

the brain's corrections would propagate to the artwork and result in in-

correct depictions of the subject matter (e.g. in a painting of a white

ﬂower in a blue-lit room, the ﬂower would look whiter than in real

life). How do representational artists learn to bypass these seemingly

automatic inferential processes? Do their brains reorganize so as to per-

ceive the true physical properties of stimuli directly, or do artists use

other cognitive strategies such as correcting inaccuracies in an artwork

by comparing the actual stimulus with an initial attemptat its represen-

tation? Previous studies have presented conﬂicting ﬁndings in this re-

gard. Graham and Meng (2011) reported that professional painters

were less susceptible than non-artists to the Craik–O'Brien–Cornsweet

effect, suggesting that these artists' direct perceptual experience of lu-

minance had changed as a result of training and practice. However,

Perdreau and Cavanagh (2011) found no differences between the abili-

ties of visual artists and non-artists to overcome luminance and size

constancy operations. Drawing ability, rather than artistic ability more

generally, has been shown to affect size constancy processes

(Ostrofsky et al., 2012), integration of object information (Perdreau

and Cavanagh, 2013), and encoding of object structure (Perdreau and

Cavanagh, 2014). Several other scholars have argued both for and

against differences between the perceptual abilities of artists and non-

artists (Fry, 1920; Gombrich, 1960; Kozbelt and Seeley, 2007; Ruskin,

1857; Thouless, 1932). How representational artists can create faithful

depictions of environments that are ﬁltered through perception is

therefore still an open question.

Perception-to-action

No matter the style or medium in which they work, artists must de-

velop the ability to translate thoughts and perceptual experiences into

skilled actions; in this case, drawing and painting. Here we conceive of

perception-to-action as encompassing those cognitive processes that

involve close interactions between perceptual and motor processes,

broadly deﬁned. Relevant perceptual processes could include visualper-

ception of the subject of an art work or of the artist's own hand, or pro-

prioceptive feedback from hand and arm as a drawing is created.

Relevant motor processes could include both the hand and arm move-

ments that create the art work and the eye movements that direct at-

tention over the subject.

Previous studies have investigated how the drawing habits and abil-

ities of artists differ from those of non-artists. Kozbelt (2001) found that

artists' superior skills in perception, motor actions, and perceptuomotor

integration contribute to their advantage in drawing ability. His data in-

dicated additionally that the perceptual advantages among artists had

developed largely to serve drawing skills. Providing further evidence

for a tight integration between perception and action among artists,

Cohen (2005) found that artists shift their eyes between the drawing

and its subject more frequently and that this gaze frequency correlates

with drawing accuracy. He suggested that frequent eye gaze shifts to

update the contents of perception allow artists to reduce the amount

of (possibly inaccurate) information held in working memory. Glazek

(2012) found additionally that artists engage more efﬁcient visual

encoding and motor output mechanisms when drawing. These results

may also relate to Perdreau and Cavanagh's (2011) argument that art-

ists overcome constancy operations essentially by trial and error: draw-

ing a subject and then making corrections after comparing the drawing

to the subject. If this is the case, one aspect of becoming skilled in draw-

ing may be development of the ability to compare drawing and subject

while creating an artwork.

An inﬂuential model of the visual system proposes that visual infor-

mation processing can be divided primarily along two neural pathways

or processing streams (Goodale and Milner, 1992; Mishkin and

Ungerleider, 1982). The ventral or vision for perception stream is respon-

sible for recovering information about object identity and tracks

features such as size, shape, and color. The dorsal or vision for action

stream is responsible for spatial awareness and the guidance of move-

ments suchas the strokes of a paintbrush. Since artists' perceptual skills

exist to subserve skilled movements, it is possible that a representation-

al visual artist's training can target the functions of the vision for action

stream over those of the vision for perception stream. If so, ﬁnding

evidence of differences between the perceptual skills of artists and

non-artists may depend on whether the tests of those skills target purely

perceptual or perception-to-action pathways.

In the current study we investigated behavioral and neural changes

in creative cognition, visual perception,and perception-to-action as fol-

lows. First, in behavioral sessions at the beginning and end of the study,

participants completed the Torrance Tests of Creative Thinking Figural

Form A (TTCT; Fig. 1B) (Torrance, 1969). The TTCT tests for the ability

to think creatively as deﬁned by several factors such as ﬂuency, original-

ity, abstractness of thought, the ability to depict complex systems com-

pellingly, and the creative use of imagery and language. The test yields a

single composite creativity index (CI) as well as submeasures that sep-

arately assess many aspects of creative ability. Although the painting

and drawing courses completed by the participants were not designed

explicitly to improve the creative qualities measured by the TTCT, we

hypothesized that training in painting and drawing would transfer to

improvements in some or all of these qualities. Because of the ﬁnding

of Jung et al. (2010) we hypothesized additionally that any changes

we observed in the creative thinking abilities of our experimental

group would correlate with corresponding changes in prefrontal white

matter organization.

Second, in order to track the development of perceptual abilities in

our visual art students and test whether improvements in these abilities

entail changes in the activity of the brain's perceptual pathways, we ac-

quired a series of functional scans in each session while participants

judged properties of illusory visual stimuli. For illusory stimuli we

used the Craik–O'Brien–Cornsweet illusion (Fig. 1C) and the Müller–

Lyer illusion (Fig. 1D). We chose these two classic visual illusions be-

cause they are cognitively impenetrable. Previous neuroimaging studies

have shown that brain areas implicated in early and mid-level visual

processing underlie the illusory effects (Perna et al., 2005; Plewan

et al., 2012). If perceived strengths and neural correlates of these visual

illusions change as students become artists, it would suggest neural

plasticity at perceptual processing levels. However, no changes in earlier

levels of processing would suggest that artists may have learned to in-

terpret the outputs of early processing differently.

Finally, to assess learning-related changes in perception-to-action

pathways, we acquired a functional scan during each session in which

participants made quick, 30 second gesture drawings based on observa-

tion of human ﬁgures (Fig. 1E). Gesture drawing is a technique often

used among representational artists to develop more direct translation

of visual observation to hand and arm movements. In creating gesture

drawings, one is often discouraged from devoting attention to the art

work itself, focusing more on translating directly from the perceived

form and gesture to motor actions that faithfully capture those aspects

of the subject on the canvas. This was especially true in the current

study, since participants lay with their heads still in the scanner and

had little opportunity to see the drawings as they were produced. If vi-

sual art training targets perception-to-action, we hypothesized that

changes in neural activity would be observed in the corresponding

perceptuomotor pathways among our art students.

Until recently, plastic reorganization of the brain was thought to

occur mainly during childhood and adolescence, leaving adults with

limited means to learn new skills. Research in the last two decades

has convincingly overturned this belief, revealing a brain that remains

able to reorganize with learning well beyond early developmental pe-

riods (Draganski et al., 2004; Lövdén et al., 2010; May, 2011; Schlegel

442 A. Schlegel et al. / NeuroImage 105 (2015) 440–451

et al., 2012; Scholz et al., 2009; Taubert et al., 2010). Structural changes

in the adult brain have been observed with interventions lasting aslittle

as six weeks (Draganski et al., 2004). Our previous work has shown that

novel insights can be gained into the neural processes underlying spe-

ciﬁc behaviors by studying howthe adult brain changes as thosebehav-

iors are learned (Schlegel et al., 2012). Thus, studying how the brain

changes as students become artists may reveal insights into how the

brain mediates artistic work.

Materials and methods

Participants

Prior to participating, 45 participants (26 females, aged 19–22 years)

with normal or corrected-to-normal visual acuity gave informed writ-

ten consentaccording to the guidelines of Dartmouth College's Commit-

tee for the Protection of Human Subjects. Data from 10 participants who

withdrew before completion of the study were discarded before further

analysis. Our ﬁnal study cohort consisted ofan experimental group of 17

undergraduate students (13 females, aged 18–22 years) who completed

a three-month-long course in either introductory drawing or introduc-

tory painting, and a control group of 18 undergraduate students (9 fe-

males, aged 19–22 years) who did not study art. Participants were

recruited for the study during the ﬁrst week of courses, and we there-

fore did not control the group assignment. Control participants were re-

cruited from a group of students taking an introductory organic

chemistry course. Our strategy was to choose a control group that en-

gaged in an equally intensive course of study, but within a system that

required closed rather than open-ended solutions. While no signiﬁcant

differences in gender, age, handedness, or grade point average existed

between the two groups (Table S1), we were not able to precisely

match these characteristics because of the nature of the group

assignment.

Drawing & painting courses

Experimental participants completed either an introductory course

in observational drawing, or a similar course in painting for which the

drawing course was a prerequisite. Both courses were 11 week,

undergraduate-level classes offered for credit through the Studio Art

Department at Dartmouth College. Classes met for 4 h each week,

with approximately 15–20 additional hours spent on work per week

outside of class. The purpose of the introductory drawing course was

to develop technical and expressive drawing abilities in terms of line,

scale, space, light, and composition. Work was primarily observational,

although non-observational drawing was also explored. The painting

course further developed the topics from the drawing course, and also

addressed paint application and color theory, mixing, and composition.

Both courses focused additionally on the development of artistic expres-

sion and on the critical analysis of art works. Detailed course descrip-

tions are provided for both courses as supplemental text.

Procedure

Participation consisted of two half-hour-long behavioral sessions,

one at the beginning and one at the end of the study, during which par-

ticipants completed the Torrance Tests of Creative Thinking (TTCT) Fig-

ural Form A, and four one-hour-long magnetic resonance imaging

(MRI) sessions, one per month, during which we collected several func-

tional and structural scans. Functional data were collected as partici-

pants performed two types of behavioral tasks. In the ﬁrst task,

participants made judgments about the physical characteristics of visual

illusion stimuli. In the second task, participants produced gesture draw-

ings basedon observation of photographs of human models. These tasks

are described in further detail below. DTI data were collected as partic-

ipants rested but lay still in the scanner.

TTCT task

The TTCT Figural Form A is a timed 30 min, standardized pencil and

paper test of creative thinking. The test consists of three 10 minute ac-

tivities in which participants are instructed to complete partially

drawn ﬁgures by thinking of things no one else would think of and by

telling as creative and elaborate a story as possible (e.g. Fig. 1B). Partic-

ipants took the TTCT twice, once at the beginning and once atthe end of

the study.

Illusion task

Stimuli were modeled after the Craik–O'Brien–Cornsweet illusion

(Fig. 1C) and the Müller–Lyer illusion (Fig. 1D), with actual differences

introduced in luminance and length, respectively. In a given trial, an il-

lusory stimulus was presented for 1 s and the participant had 2 s from

stimulus onset to indicate which side was either brighter (for the

Craik–O'Brien–Cornsweet or “luminance”task)orlonger(fortheMüll-

er–Lyer or “length”task) before the next stimulus appeared. The Craik–

O'Brien–Cornsweet illusion stimulus measured 22.72° of visual angle

wide and 17.14° tall, while each center gradient was 3° wide. The Müll-

er–Lyer line segments were oriented vertically and centered on the

screen, with a 4° horizontal separation between them. Line segments

ranged in length from 4° to 8°. The gradient directions and lighter side

for the luminance task and the arrow end direction and longer side for

the length task were counterbalanced across trials, and participants

were not given feedback on their responses. During each session, partic-

ipants completed one fMRI run of each illusion task. A run consisted of

ﬁve blocks of 20 trials each, yielding 100 trials per task, participant,

and session. Blocks were interleaved with 12 second rest periods.

Gesture drawing task

Stimuli consisted of grayscale photos of human ﬁgures in various

poses (e.g. Fig. 1E). Before the ﬁrst scan, participants were shown exam-

ples of gesture drawings and instructed that gesture drawing is an activ-

ity in which ﬂuid, continuous strokes are used to build up the form and

movement of the observed ﬁgure quickly rather than to capture ﬁne de-

tails. In each trial, a ﬁgure was shown to the participant who then had

30 s to complete a gesture drawing of that ﬁgure. Participants used a

#2 pencil while lying supine in the scanner and resting a wedge-

shaped drawing pad and paper on their abdomens. Because of their po-

sition and the instruction to keep their head still during scans, partici-

pants had little opportunity to view the gesture drawings as they

were produced. This constraint was not a major concern because, as

discussed above, gesture drawing technique stresses that attention be

maintained on the subject rather than the art work. An experimenter

stood at the side during the scan to provide a new piece of paper in be-

tween drawings. During each session, participants completed one fMRI

gesture drawing run, consisting of 10 drawings interleaved with 14 sec-

ond rest periods.

MRI acquisition

MRI data were collected using a 3.0-Tesla Philips Achieva Intera

scanner with a 32-channel sense head coil located at the Dartmouth

Brain Imaging Center. T2*-weighted gradient echo planar imaging

scans were used to acquire functional images covering the

whole brain (2000 ms TR, 20 ms TE; 90° ﬂip angle, 240 × 240 mm FOV;

3×3×3.5mm

voxel size; 0 mm slice gap; 35 slices). One T1-weighted

structural image was collected each session using a magnetization-

prepared rapid acquisition gradient echo sequence (8.176 ms TR;

3.72 ms TE; 8° ﬂip angle; 240 × 220 mm

FOV; 188 sagittal slices;

0.9375 × 0.9375 × 1 mm

voxel size; 3.12 min acquisition time). One

diffusion-weighted scan was collected each session to acquire diffusion

tensor imaging (DTI) data (32 directions; 1000 s/mm

b-value; 8.379

443A. Schlegel et al. / NeuroImage 105 (2015) 440–451

s TR; 73.49 ms TE; 90° ﬂip angle; 224 × 224 mm

FOV; 70 axial slices;

2×2×2mm

voxel size; one additional b = 0 s/mm

image; 2 signal

averages per volume; 10.47 min acquisition time).

MRI data preprocessing

fMRI and DTI data were preprocessed using FSL (Smith et al., 2004).

Functional data were motion and slice scan time corrected, temporally

high pass ﬁltered with a 1/100 Hz cutoff, and spatially smoothed with

a 6 mm full-width-at-half-maximum Gaussian kernel. For each DTI

data set, fractional anisotropy (FA), radial diffusivity (RD), axial diffusiv-

ity (AD), and mean diffusivity (MD) were reconstructed, z-transformed

to reduce the effect of potentialunexpected changesin equipment func-

tion over time, and normalized to a standard 2 × 2 × 2 mm

template in

Montreal Neurological Institute (MNI) space via FSL's FNIRT non-linear

registration tool. High-resolution anatomical images were processed

using the FreeSurfer image analysis suite (Dale et al., 1999).

DTI data analyses

Our DTI data analyses followed the procedure from Schlegel et al.

(2012).

Whole-brain GLM

A cross-subject longitudinal GLM analysis using FSL's randomise tool

was performed on the whole-brain FA data (5000 permutations,

within-subject only). Data from all four sessions were included in the

analysis. Only data from white matter were analyzed,as determined in-

dividually for each participant by FreeSurfer's automatic segmentation

algorithm. Additionally, each voxel location in MNI space was only in-

cluded in the analysis if at least 75% of participants had white matter

at that location. Even for locations that passed the 75% threshold, data

from non-white matter at those locations were excluded from the anal-

ysis. This process allowed for as much of the white matter data as pos-

sible to be included without risking errors due to potential

misregistration between white and non-white matter regions.Included

in the GLM design matrix were predictors to account for existing differ-

ences between participants and predictors for time, separated by group.

t-contrasts were deﬁned to assess group by time interaction effects, and

the resulting statistical maps were threshold-free cluster enhanced and

false discovery rate (FDR) corrected for multiple comparisons (Smith

and Nichols, 2009). The same analysis was also performed for RD, AD,

and MD measures.

Time course

For each participant, the same GLM as in the whole-brain longitudi-

nal DTI analysis was performed with that participant's data left out. The

participant's FA data at voxels with p≤0.05 (uncorrected) in the group

by time interaction contrast of this analysis were then averaged to com-

pute one mean FA value for that participant at each time point. Uncor-

rected pvalues were used in this voxel selection procedure to

maximize the amount of meaningful data included in the time courses,

while excluding the data to be averaged from the voxel selection step

prevented type I errors in the GLM analysis from propagating to the

time courses (i.e. decreasing the slope of the participant group and in-

creasing the slope of the control group artiﬁcially). Results for each par-

ticipant were converted to percent change from the t = 0 value of a

best-ﬁt line for that participant's data, and these time courses were

then averaged by group and shown in Fig. 3B.

Structural MRI data analysis

The same whole-brain longitudinal GLM analysis as described for

the DTI data analyses was performed on gray matter thickness data as

reconstructed by FreeSurfer.

TTCT data analyses

Behavioral

The TTCTs were scored independently and blindly according to

standard rating procedures (see Torrance, 1969) by two indepen-

dent raters who also evaluated the gesture drawings (described

below). Inter-rater reliability was excellent (α=0.94forthecom-

posite creative index [CI]). The group by time interaction term of a

two-way, repeated measures ANOVA evaluated whether the change

in CI over the study differed between the two groups. A factor analysis

was performed on all TTCT subscores including a modiﬁcation of the

checklist of creative strengths in which the total number of instances of

each checklist item was recorded for each test section. Factors were ex-

tracted based on principal components with eigenvalues greater than 1

and varimax rotated (resulting in the ﬁve factors shown in Fig. 2B). Ad-

ditional one-tailed, unpaired t-tests evaluated whether each of these

factors increased more in the experimental group than the control

group over the study period. p-Values were FDR corrected for the ﬁve

comparisons.

Correlation with DTI data

For each identiﬁed TTCT factor, a between-subject correlation analy-

sis was performed comparing the change in factor score between the

two test administrations and the slope of the FA changes calculated

above. For Factor 2, which showed a signiﬁcant result in this analysis,

the same correlation analyses were performed between FA slope and

the change in the four TTCT subscores that loaded highly (coefﬁ-

cient ≥0.4) onto this factor. The resulting p-values were FDR corrected

for multiple comparisons across the four submeasures.

Illusion data analyses

Behavioral

We reconstructed a psychometric c urve for each fMRI illusion run by

ﬁtting a Weibull function to participant responses for that run. Re-

sponses were coded for the Craik–O'Brien–Cornsweet luminance task

based on whether the participant chose the side with the lighter or

darker gradient and for the Müller–Lyer length task based on whether

the participant chose the arrow with the outward or inward facing

ends. The tparameter from the Weibull ﬁt gives an estimate of the

point of subjective equality (or the actual luminance or length differ-

ence that led the participant to make each choice with equal probabili-

ty) and hence the strength of the illusory effect. The bparameter from

the ﬁt determines the steepness of the psychometric curve and thus in-

dicates the consistency of the participant's decision criterion. The group

by time interaction terms from longitudinal linear mixed model (LMM)

analyses were used to determine whether the trajectory of these two

parameters differed between the two groups over the study period.

fMRI univariate

A whole-brain longitudinal GLM analysis was carried out for each il-

lusion task to determine whether brain activity changed differentially

between the two groups during the study. Only data from cortical

gray matter and subcortical structures were analyzed, as determined

by FreeSurfer's automatic segmentation algorithm (Dale et al., 1999).

Aﬁrst-level GLM in FSL's FEAT tool for each participant and session

used a boxcar predictor to model differences in blood-oxygenation

level dependent (BOLD) measurements between task and rest, con-

volved with a double-gamma hemodynamic response function (Smith

et al., 2004). The model parameter estimates from this analysis were

passed to a higher-level longitudinal GLManalysis using FSL's randomise

tool (5000permutations, within-subject only), like that used for the DTI

analysis described above. t-Contrasts were deﬁned to assess group by

time interaction effects, and the resulting statistical maps were

threshold-free cluster enhanced and FDR corrected for multiple

comparisons.

444 A. Schlegel et al. / NeuroImage 105 (2015) 440–451

fMRI multivariate

Multivariate pattern classiﬁcations were carried out using PyMVPA

(Hanke et al., 2009;seeNorman et al., 2006 for a detailed introduction

to multivariate classiﬁcation analysis). For each session we performed

a whole-brain, cross-subject multivariate classiﬁcation analysis, in

which we trained a machine classiﬁer to distinguish between experi-

mental and control participants based on patterns of brain activity

measured during the gesture drawing task. We used a whole-brain clas-

siﬁcation procedure rather than a searchlight analysis for two reasons.

First, searchlight analyses are only sensitive to localized patterns of in-

formation, since each classiﬁcation is performed only on a localized

cluster of voxels. A whole-brain classiﬁcation allows patterns to be dis-

tributed across the brain, as might be expected for complex cognitive

processes. Second, searchlight analyses entail a large number of multi-

ple comparisons that can be avoided by performing a single whole-

brain classiﬁcation. As classiﬁcation data we used the parameter esti-

mates (i.e. “beta values”)fromtheﬁrst-level univariate analysis de-

scribed above. These parameter estimates represent the change in

brain activity in each voxel while drawing, compared to rest. During

classiﬁcation, the machine classiﬁer is ﬁrst trained by providing it with

a set of patterns along with the associated labels (in this case “experi-

mental”or “control”). After training, the classiﬁer is tested by giving it

a new pattern that was held out of the training step and determining

whether it can assign the correctlabel to that pattern. This training/test-

ing step is repeated in a cross-validation procedure in which each pat-

tern is successively held out of the training data and included in the

testing data. If the classiﬁer can label the test patterns correctly above

chance over all of the cross-validation folds, then information exists

within the patterns that distinguishes the two groups. In a whole-

brain classiﬁcation, an initial voxel selection procedure is conducted to

determine a subset of voxels in the whole cortex to include in the

classiﬁcation. Speciﬁcally, in each fold of the cross-validation we ﬁrst

performed a one-way ANOVA between the two groups' parameter

estimates (leaving out data from the held out participant). The 1% of

voxels with the highest F-scores in this ANOVA were then selected for

training and testing. This selection procedure allowed us to reduce the

dimensionality of the multivariate patterns in a principled way while

not biasing the classiﬁcation results, since the testing data were left

out of the selection mechanism. One limitation of the whole brain

classiﬁcation selection procedure is that it may discard voxels that still

contain information that distinguishes the two groups, but not enough

to pass the selection threshold (i.e. not showing large enough between

group differences). The selection threshold we chose, however, allowed

us to strike a balance between ﬁnding information if it was present and

avoiding overﬁtting of the data. We used a linear support vector ma-

chine (SVM) as our classiﬁcation algorithm, and leave-one-subject-out

cross validation (i.e. one participant's data were held out from the train-

ing set in each cross-validation fold). Because the training data were un-

balanced between groups, we included an additional step in which the

classiﬁer was trained and tested 10 times for each cross-validation

fold, using a random subset of balanced data for training. The classiﬁer

performance for each fold was then calculated as the average perfor-

mance across those 10 subfolds. Classiﬁcation accuracies from these

analyses were compared to chance using one-tailed Monte Carlo tests

with 1000 permutations.

Gesture drawing data analyses

Behavioral

All gesture drawings were digitized and scored independently and

blindly by two trained raters. One rater was a visual artist who was

not familiar with the participants' artwork and the other was a research

assistant who was trained by the ﬁrst rater. The score for each drawing

was an integer between 1 (lowest quality) and 10 (highest quality),

based on an assessment of line quality, resemblance to the ﬁgure, and

expressiveness of the depicted gestures. The inter-rater reliability was

excellent (α= 0.82). The group by time interaction effect from an

LMM analysis was used to determine whether the change in these rat-

ings differed between the two groups over the study period.

fMRI

To evaluate whether differences in head movement could have led

to our observed effects, we conducted two-tailed, unpaired t-tests

Fig. 2. Results of creative thinking analysis. Asterisks indicate signiﬁcance. Errorbars indi-

cate standard errors of the mean. A. Art students increased in creative thinking ability

compared to the controls, as measured by the TTCT composite creative index. B. A factor

analysis of the 18 TTCT submeasures revealed ﬁve factors that accounted cumulatively

for 59.7% of variance in the participant responses. Factor loadings for each submeasure

are shown, multiplied by 10 for presentation. Loadings with magnitude greater than 0.4

are highlighted andcolor coded by increasing magnitude (red to yellow). C. Scores for Fac-

tors 1 through 3increased in art students relative to controls, indicating that art students

improved in their ability to think divergently, model systems and processes, and portray

imagery. p-Values are FDR corrected for the ﬁve comparisons.

445A. Schlegel et al. / NeuroImage 105 (2015) 440–451

between the two groups for each session using the per-run mean rela-

tive displacement as reported by FSL's MCFLIRT tool. Univariate and

multivariate analyses were carried out on the gesture drawing fMRI

data as in the illusion analyses described above. To conﬁrm that the

above-chance multivariate classiﬁcation accuracies observed were not

due to gross BOLD level differences between the two groups, an addi-

tional higher-level univariate whole-brain GLM analysis wasperformed

for each session using randomise (5000 permutations, threshold-free

cluster enhanced and FDR corrected for multiple comparisons), with t-

contrasts deﬁned to evaluate between-group BOLD differences. As an

additional control to conﬁrm that the effects we observed were due to

ﬁne-scale patterns of activity, the same multivariate analysis was per-

formed using the mean pattern values as samples rather than the pat-

terns themselves.

Results

Creative cognition

In behavioral sessions at the beginning and end of the three month

study period, participants completed the TTCT Figural Form A, a test of

conceptual creativity. All tests were scored independently and blindly

by two trained raters. The inter-rater reliability was excellent (α=

0.94). For each participant, we calculated the change between the two

sessions in the composite creative index (CI) given by the TTCT

(Fig. 2A). A two-way, repeated measures ANOVA comparing the abso-

lute TTCT CI scores showed a signiﬁcant interaction between group

and time, indicating that the art students' creative thinking ability im-

proved signiﬁcantly compared to control participants during the study

(F(1,32) = 4.47, p= 0.0423). Note that differences between the two

groups' CI measures existed at the ﬁrst session (t(32) = −3.85, pb

0.001; see Table S2). Because creativity is a complex trait (Dietrich

and Kanso, 2010; Hee Kim, 2006),reducing creativecognition to a single

quantity such as CI would limit our understanding of the changes that

occurred. CI is actually derived from 18 different submeasures of crea-

tivity (see Fig. 2B), so we performed a factor analysis on these

submeasures in order to extract multiple dimensions along which to ex-

amine the changes that occurred among our participants. This analysis

yielded ﬁve factors that together explained 59.7% of the variance in

the TTCT data (Fig. 2B). The submeasures ﬂuency,originality,breaking

boundaries,elaboration, and resisting premature closure loaded highly

onto Factor 1. These submeasures all target divergent thinking, deﬁned

as the ability to produce many (ﬂuency,elaboration)original(originality,

breaking boundaries,resisting premature closure) solutions to a problem

(Guilford, 1967). Storytelling articulateness,synthesis of ﬁgures,depicting

movement or action, and emotional expressiveness loaded highly onto

Factor 2. These submeasures all required the modeling of a system or

process (story narratives, the continuation of a theme over multiple

panels, movements and actions, and emotional states). Richness of imag-

ery,colorfulness of imagery,andelaboration loaded highly onto Factor 3

and shared a common theme of requiring the depiction of rich, complex,

effective imagery. Expressiveness of titles,abstractnessof titles,andhumor

loaded highly onto Factor 4, withthe common theme ofinvolving verbal

creativity. And originality and synthesis of lines loaded highly onto Factor

5, with no discernible interpretation.

We calculated the change in each of these ﬁve factor scores for each

participant and performed one-tailed, unpaired t-tests between the ex-

perimental and control groups to determine which factors were respon-

sible for the improvements in creative thinking among the art students.

The results of this analysis showed that art students improved over con-

trols in the creative qualities captured by Factors 1, 2, and 3 (divergent

thinking, modeling of systems and processes, and imagery; Fig. 2C;

Table S3). Factors 4 and 5 showed no signiﬁcant changes between the

two participant groups.

In order to assess whether these improvements in creative ability re-

lated to changes in white matter structure, we ﬁrst performed a whole-

brain (white matter only), permutation-based longitudinal GLM analy-

sis on the Fisher's Z transformed fractional anisotropy (FA) values de-

rived from the DTI scans collected for each participant and session. FA

quantiﬁes the degree to which the diffusion of water is biased in partic-

ular directions andis associated withvariations in axoncount and diam-

eter, density of axonal packing, myelination, and the coherence of axon

ﬁber bundles (Beaulieu, 2011). This analysis revealed bilateral, prefron-

tal white matter clusters in which FA decreased in art students relative

to controls over the course of the study (Fig. 3A; all ps FDR corrected for

multiple comparisons). No other signiﬁcant effects of training were

found (no voxels were signiﬁcant after FDR correction) when the

same analysis was applied to radial diffusivity (amount of diffusion per-

pendicular to the principal axis), axial diffusivity (amount of diffusion

parallel to the principal axis), mean diffusivity (mean amount of diffu-

sion), or cortical gray matter thickness instead of FA.

To conﬁrm that the clusters identiﬁed in the longitudinal FAanalysis

were associated with absolute decreases in FA in the art students rather

than increases in FA in the control group, we calculated the mean per-

cent change in FA within these clusters for each participant and session

in both groups. However, calculating FA time courses directly from the

voxels that reached signiﬁcance in the longitudinal analysis would de-

crease the slope of the experimental group's time course and increase

the slope of the control group's time course artiﬁcially due to noise-

induced false positives in the whole-brain analysis. In order to prevent

these false-positives from affecting the result, for each participant we

calculated the mean FA percent change at each time point only in voxels

that were signiﬁcant in the same longitudinal GLM analysis conducted

with that participant's data left out (cf. Schlegel et al., 2012). An LMM

analysis of these FA change time courses revealed that art students ex-

perienced a progressivedecrease in FA in bilateralprefrontal white mat-

ter (group by time interaction term t=−2.41, p= 0.016; Fig. 3B).

To evaluate whether a relationship existed between the visual art

students' decreases in prefrontal FA and increases in creative ability,

we performed correlation analyses between the FA percent change cal-

culations and the changes in each factor score between the two admin-

istrations of the TTCT,with results shownin Fig. 4A. After FDR correction

for multiple comparisons across the ﬁve factors, onlychanges in Factor 2

showed a signiﬁcant inverse correlation with FA percent change. There-

fore, decreases in FA in prefrontal white matter were associated with in-

creases in the ability to model systems and processes creatively.

Additional correlation analyses with the four TTCT submeasures that

loaded highly onto Factor 2 revealed thatchanges in three of these –sto-

rytelling articulateness, portraying movement or action, and emotional

expressiveness –showed an inverse relationship with changes in pre-

frontal white matter FA (Fig. 4B; all ps FDR corrected for multiple com-

parisons across the four submeasures). Changes in the fourth

submeasure –synthesis of ﬁgures –showed no signiﬁcant correlation

with changes in prefrontal FA.

Visual perception

During each scanning session, participants completed two block-

design runs in which they made judgments about two types of illusory

visual stimuli. In the ﬁrst “luminance”run, stimuli were based on the

Craik–O'Brien–Cornsweet illusion (Fig. 1C). In this illusion, a dark-to-

light gradient and a light-to-dark gradient heading in opposite direc-

tions from the center of a uniform gray rectangle cause the rectangle

to appear darker on one side than the other. We introduced actual dif-

ferences in the luminance between the two sides of the rectangle and

asked participants to judge which side was lighter. In the second

“length”run, stimuli were based on the Müller–Lyer illusion (Fig. 1D),

in which a line segment with outward facing arrow ends appears longer

than an identical line segmentwith inward facingarrow ends. We intro-

duced actual differences in the length of the two line segments and

asked participants to judge which line segment was longer. We then es-

timated psychometric response curves for each task, session, and

446 A. Schlegel et al. / NeuroImage 105 (2015) 440–451

participant in order to assess a) the point of subjective equality and thus

strength of the illusory effect and b) the steepness of the psychometric

curve and thus the consistency of the participants' decision criteria.

Using these data, we performed an LMM analysis to evaluate whether

either measure changed differentially between the two groups during

the study and found no effect in either measure or task (Fig. S1 &

Table S4; all psN0.05, uncorrected). Although we found no enhance-

ment in perceptual ability in the experimental group, we were still curi-

ous whether training would lead to differences in brain activity while

making perceptual judgments. To address this question, we ﬁrst per-

formed whole-brain univariate longitudinal analyses on the BOLD mea-

surements from each of the illusion tasks. For each task we used a

permutation-based GLM with t-contrasts deﬁned to assess group by

time interactions in BOLDactivity, and found no signiﬁcant effects for ei-

ther task (all psN0.05, FDR corrected for multiple comparisons). While

there were nounivariate BOLD differences between the groups, it is still

possible that differences in more ﬁne-grained patterns of activity arose

between the groups as a result of the art students' training. To assess

this possibility, we performed a whole-brain multivariate classiﬁcation

analysis for each session and illusion task in which we trained a linear

SVM to classify between the experimental and control groups based

on patternsof BOLD activity (Norman et al., 2006). If pattern level differ-

ences did arise as the art students learned, we would expect classiﬁca-

tion accuracy to increase progressively across the sessions. In order to

reduce the dimensionality of the classiﬁcation patterns, we used an ini-

tial voxel selection procedure to choose the 1% of voxels that showed

the highest F-scores in a one-way ANOVA between the two groups' pa-

rameter estimates in a GLM contrast of task vs. rest, excluding the cross-

validation holdout sample. None of these classiﬁcation analyses

achieved accuracies signiﬁcantly above chance (one-tailed Monte

Carlo test,Fig. S2 & Table S5; all psN0.05, uncorrected). The lack of effect

in any of these analyses supports our behavioral ﬁndingthat no changes

occurred between the two groups in these tasks and suggests that the

art training did not lead to changes in perceptual skill or strategy in

the perceptual judgment tests that we used.

Perception-to-action

In each scanning session, participants also completed one block-

design run in which theycreated 30 second gesture drawings while ob-

serving grayscale images of human ﬁgures in various poses (Fig. 1E).

Each drawing was scored independently and blindly by two trained

raters based on an assessment of line quality, resemblance to the ob-

served ﬁgure, and recognizability and expressiveness of the depicted

gestures. The inter-rater reliability was excellent (α= 0.82). The

drawing ratings were averaged for each participant and session. A

two-tailed, unpaired t-test found no signiﬁcant difference between the

two groups' drawing skills in the ﬁrst session (t(33) = 1.56, p=

0.128). However, an LMM analysis performed on these mean ratings

conﬁrmed that the art students improved over time in gesture drawing

ability relative to the controls (group by time interaction t= 3.06, p=

0.003; Fig. 5A).

Thus, the art students improvedin both creative cognitive ability and

in technical skills related to gesture drawing. To evaluate whether these

two areas of improvement were related, we performed a correlation

analysis between the change in TTCT CI between our two test adminis-

trations and the slope of our experimental participants' gesture drawing

ability over the four neuroimaging sessions. We found no signiﬁcant re-

lationship between change in creativity and change in gesture drawing

ability (r(14) = 0.0606, p= 0.412). This null ﬁnding underscores the

separation between creative and t echnical abilities and the m ultifaceted

nature of the training program undertaken by the art students.

Next, we asked whether the experimental group's enhancement in

drawing ability tracked changes in gesture drawing related brain activ-

ity over the course of the study. To control for the possibility that differ-

ences in drawing-induced head movements between the two groups

could have led to differences in BOLD measurements, we performed

per-session two-tailed, unpaired t-tests on our fMRI motion correction

mean relative displacement measurements. During no session was

there a signiﬁcant difference in head movement between the two

groups (all psN0.05, uncorrected; Table S6). We performed whole-

brain univariate longitudinal and whole-brain multivariate cross-

subject classiﬁcation analyses as described above for the illusion tasks.

As in the illusion tasks, no signiﬁcant univariate differences in brain ac-

tivity developed over time between the two groups (all psN0.05, FDR

corrected for multiplecomparisons). This suggests thatno mean activity

differences between participant groups developed over the course of

the study, as might be expected if, for instance, art students drew

more vigorously as they underwent training. However, we did ﬁnd

that the SVM classiﬁer in our whole-brain multivariate analysis distin-

guished more accurately between art students and controls based on

drawing-related brain activity as the study progressed (Fig. 5B;

Table S7). Initially, the classiﬁer failed to distinguish signiﬁcantly be-

tween the groups (p= 0.249 in a one-tailed Monte Carlo test compared

to chance),but by the end of the study the classiﬁer achieved 82.9% clas-

siﬁcation accuracy (p= 0.002 in the same test). To determine which

areas of the brain showed patterns of activity that supported the classi-

ﬁcation between art students andcontrols, we visualized the voxels that

were selected by our voxel selection procedure in at least 50% of classi-

ﬁcation folds (Fig. 5C). Although clusters of voxels were selected

Fig. 3. Results of longitudinal analysis of white matter structure. Error bars indicate standard errorsof the mean. A. Voxels that showed a progressive decrease in Fisher's Z transformed

fractional anisotropy(FA) in art students relative to controls, shown within a glass brainin MNI space (p≤0.05, FDR corrected for multiple comparisons). All voxels are inwhite matter,

and the majority of voxels arewithin the frontal lobe.B. Time course of FA in signiﬁcant voxelsfrom the analysisin panel A. The art students' prefrontal FA decreased progressively overthe

study period.

447A. Schlegel et al. / NeuroImage 105 (2015) 440–451

throughout the brain, with some clusters being chosen presumably due

to noise, thelargest cluster selected in each month occurred consistent-

ly in the right hemisphere of the cerebellum. Buckner et al. (2011) dem-

onstrated previously that regions of the cerebellum overlapping with

those found in the present analysis project to the hand and arm regions

of left primary motor cortex and are involved in proprioceptive feed-

back. Miall et al. (2001) found evidence that similarly overlapping re-

gions are responsible for coordinating eye and hand movements.

Subsequent work has conﬁrmed their ﬁndings and found that these cer-

ebellar regions change with visuomotor task learning (Floyer-Lea and

Matthews, 2004; Maquet and Schwartz, 2003). Interestingly, the selec-

tion procedure chose clusters of voxels consistently in the left motor

cortex as well (see Fig. 5C). These chosen voxel clusters suggest that

the classiﬁcation analysis picked up on patterns of complex hand- and

arm-related motor activity. To conﬁrm that these signiﬁcant classiﬁca-

tions did not occur due to gross BOLD level differences in activity

between the groups, for each session we performed a whole-brain uni-

variate GLM analysis to assess differences in brain activity between art

students and controls. No voxels in any of these analyses reached signif-

icance (all psN0.05, FDR corrected for multiple comparisons). As a fur-

ther control to test whether mean activity levels or ﬁne grained spatial

patterns of information were responsible for the signiﬁcant classiﬁca-

tion result, we tested our classiﬁer again, this time using the mean

value of each pattern rather than the pattern itself. None of these classi-

ﬁcations achieved accuracies signiﬁcantly greater than chance (all psN

0.05, uncorrected). Thus, ﬁne-scale patterns of gesture drawing related

neural activity in the cerebellum and cortex increasingly differentiated

art students from controls during the course of the study, while overall

neural activity levels remained the same. These results suggest that our

participants' training in visual art entailed a reorganization of ﬁne scale

functions in these areas.

Discussion

Here we show that the brains of young adults reorganize as they learn

to create visual art. Our controlled, longitudinal design involving training

over three months allowed us to rule out possible confounding effects of

normal aging and maturation as well as differences in motivation and ini-

tial expertise levels between the experimental and control groups. We did

not ﬁnd any improvements in the art students' purely perceptual skills or

related brain activity relative to a control group of students who did not

study art. We did, however, ﬁnd that the art students improved in the

ability to quickly translate observations of human ﬁgures into gesture

drawings and that ﬁne-grained patterns of drawing-related neural activ-

ity in the cerebellum and cerebral cortex increasingly differentiated the

art students from the control group over the course of the study. This ﬁnd-

ing complements a recent correlational study by Chamberlain et al.

(2014) showing that gray matter density in left anterior cerebellum and

right medial frontal gyrus is higher in individuals who have trained in ob-

servational drawing. From their ﬁnding, it was concluded that observa-

tional drawing ability depends on differences in structural organization

related to ﬁne motor control in these areas. However, we did not ﬁnd

that gray matter thickness changed as a function of training in these or

any other cortical areas. There are many possible interpretations of this

discrepancy, among them that Chamberlain and colleagues' cross-

sectional design could have found pre-existing differences in gray matter

organization that determined innate potentialities for differences in

drawing ability. Interestingly, the art students in our study also improved

in measures of creative thinking, speciﬁcally in their ability to think diver-

gently, model systems and processes, and use imagery. Increases in their

ability to model systems and processes creatively correlated with de-

creases in the fractional anisotropy (FA) of prefrontal white matter. Con-

sistent with a previous study that found creative ability to be negatively

correlated with prefrontal white matter FA (Jung et al., 2010), our results

suggest that changes in prefrontal white matter FA may underlie the pro-

cesses by which art training leads to increased creative ability. The lack of

correlation between our observed improvements in TTCT scores and ges-

ture drawing ability supports the view that artistic training is a multifac-

eted process involving the development of both creative cognitive and

technical skills.

Our results are also consistent with models that attribute improve-

ments in representational artists' skills to their ability to translate percep-

tion into creative action rather than to perceive the world differently (cf.

Gombrich, 1960). Like Perdreau and Cavanagh (2011), we found no spe-

cial ability among artists to extract veridical properties of the physical en-

vironment through purely perceptual processes. It should be noted,

however, that both our study and that of Perdreau and Cavanagh tested

a speciﬁc subset of perceptual abilities; it is possible that other tests

could have revealed changes in perceptual ability that our studies were

unable to observe. We did, nonetheless, ﬁnd that art students and non-

artist controls became progressively more distinguishable via drawing-

related patterns of neural activity in regions of the cerebellum and

Fig. 4. Results of correlation analysis between changes in white matter organization and

changes in creative thinking. A. Correlations between changes in factor scores (Fig. 2)

and changesin prefrontal whitematter FA (Fig. 3). Changesin Factor 2 inverselycorrelated

with changesin prefrontalFA. All ps are FDR corrected for theﬁve comparisons. B. Changes

in three of thefour TTCT submeasures that loadedhighly onto Factor2 also showed an in-

verse correlation with changes in prefrontal FA. All ps are FDR correcte d for the four

comparisons.

448 A. Schlegel et al. / NeuroImage 105 (2015) 440–451

motor cortex that have been shown previously to mediate ﬁne motor

control, proprioceptive feedback, and coordination between eye and

hand movements, and are likely involved in non-motor cognit ive process-

es as well (Buckner, 2013; Floyer-Lea and Matthews, 2004; Maquet and

Schwartz, 2003; Miall et al., 2001). Thus, our ﬁndings provide evidence

that visual art training changes neural processing in regions that mediate

integration between perception and action. These results are consistent

with the possibility that differences in at least some representational

artist's perceptual abilities may only become apparent when coupled

with action, such as the skilled strokes of a paintbrush or the building

up of a work of art over time through continuous comparison between

what is observed and what is produced. In line with this view, Kozbelt

(2001) showed that artists were better at several perceptual tasks but

also that these advantages had developed largely to support drawing

skills. Future research could help resolve these issues by examining

explicitly the source of representational artists' increased accuracy in

creating art works from observation.

The prefrontal cortex is associated with many complex behaviors that

involve long-term goal making and planning (Miller et al., 2002; Tanji and

Hoshi, 2001), generating novel and ﬂexible rules (Rougier et al., 2005),

and imagining future events (Addis et al., 2007), among other abilities.

Common to many of these skills is the ability to represent complex pro-

cesses that do not currently and may neverexistintheimmediateexter-

nal environment (Frick et al., 2014). The prefrontal cortex therefore likely

plays an important role in creative behavior and especially the creative

work of an artist, which requires the ﬂexible development of novel and

complex thoughts, processes, and objects. Indeed, Jung et al. (2010)

used DTI to ﬁnd that more creative individuals exhibited lower FA in pre-

frontal white matter tracts. However, the causal inferences that can be

made from their cross-sectional design are limited. While their results

are consistent with our ﬁndings, our longitudinal design provides addi-

tional evidence that creative cognition can improve with training on a

time scale of three months and that prefrontal white matter reorganizes

as participants become more able to think creatively.

It is worth noting that DTI measures water diffusion and so provides

only an indirect measure of white matter organization. The precise micro-

structural correlates of FA are generally difﬁcult to elucidate, although FA

is associated with several properties of individual axons and axon bundles

(Beaulieu, 2011). As creative thought often requires forming many con-

nections between disparate concepts, one possibility for the effects we

observed is that improvements in creative thinking are associated with

an increase in the complexity of axonal packing (i.e. a more complex

pattern of connectivity) in the frontal lobes. It is also possible that axonal

demyelination played a role. Evidence has been found of experience-

dependent myelination of axons in mice (Demerens et al., 1996), but a de-

crease in FA such as observed in the present study would be associated

with decreases in axonal myelination and would require that previously

unobserved learning-induced demyelination processes exist in the adult

human brain. To speculate, this demyelination could occur as art students

develop more efﬁcient processing pathways (Solso, 2001)orasmorecre-

ative individuals learn to avoid use of frontal inhibitory circuits. There is

evidence that creative activity involves the inhibition of prefrontal activity

Fig. 5. Resultsof the gesture drawinganalysis. Error bars indicate standarderrors of the mean. A. A longitudinal LMM analysis revealed that art studentsimproved progressively in gesture

drawing ability relative to controls overthe four sessions of the study.B. A whole-brain multivariate between-group linear SVM classiﬁcationanalysis based on patterns of gesture draw-

ing-related brain activity improved progressively in its ability to distinguish art students from controls. Plot shows classiﬁcation accuracy over time. Initially, the classiﬁer could not dis-

tinguish art students from controls. By the end of the study, the classiﬁer could distinguish between the groups with 82.9% accuracy (#: p≤0.1, **: p≤0.01). C. Voxels selected in each

session by the analysis in panel B to be includedin the classiﬁcation patterns, in MNI space. Voxels are included in these plots if they were selectedin at least 50% of the cross-validation

folds. The largest clusterin each session is colored green.In each session, this cluster is in the area of the right anterior lobe ofthe cerebellum that projects to the hand and arm regions of

the left motor cortex.

449A. Schlegel et al. / NeuroImage 105 (2015) 440–451

(Limb and Braun, 2008). Alternatively, development of the glial support

network resulting from increased use of the white matter tracts we iden-

tiﬁed (e.g. to actively direct creative thought) could have led to a more

complex extra-axonal environment and thus decreased FA. For instance,

oligodendrocytes extend processes that wrap myelin sheaths around

axons in an activity-dependent manner (Barres and Raff, 1993;

Demerens et al., 1996). While increases in myelination may increase FA,

extension of the orthogonally-oriented processes of oligodendrocytes

might randomize water diffusion and thereby lower FA. Evidence exists

that glial cells are modiﬁed by neuronal activity (Barres and Raff, 1993;

Ishibashi et al., 2006), but we are not aware of direct evidence that

these changes can affect FA. Future work that elucidates the cellular

basis of learning-induced changes in white matter organization could

shed further light on interpreting the results presented here.

How does the human brain mediate complex, creative processes

such as the construction of representational works of visual art? An-

swers to this and related questions will not only illuminate the artistic

process itself but could also lead to a more general understanding of

the ﬂexible behaviors that set humans apart as a species. An artist's

work is not an innate skill and so must be developed through study

and practice. The present study reveals that, at least in some cases, artis-

tic development is a complex process involving changes in behavior

along with the reorganization of both the structure and function of the

brain. While we documented changes in speciﬁc aspects of creative cog-

nition and in the integration of perception andaction, future research in

different artistic disciplines and among different populations will un-

doubtedly reveal further complexity in the creative learning process.

Acknowledgments

We thank Brenda Garand for her advice and support. This study was

funded by a National Science Foundation Graduate Research Fellowship

(No. 2012095475) to AS, Templeton Foundation Grant 23437 to PUT,

and Dartmouth Internal Funding to MM.

AS, ER, PUT, and MM designed the study. ER taught the art courses

(with other professors). AS, PA, SVF, XL, and ZL collected data. AS and

SVF developed analytical tools. AS, PA, SVF, and PJK analyzed the data.

All authors were involved in writing the paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.

doi.org/10.1016/j.neuroimage.2014.11.014.

References

Addis, D.R., Wong, A.T., Schacter, D.L., 2007. Remembering the past and imagining the future:

common and distinct neural substrates during event construction and elaboration.

Neuropsychologia 45, 1363–1377. http://dx.doi.org/10.1016/j.neuropsychologia.2006.

10.016.

Arden, R., Chavez, R.S., Grazioplene, R., Jung, R.E., 2010. Neuroimaging creativity: a psycho-

metric view. Behav. Brain Res. 214, 143–156. http://dx.doi.org/10.1016/j.bbr.2010.05.

015.

Barres, B.A., Raff, M.C., 1993. Proliferation of oligodendrocyte precursor cells depends on

electrical activity in axons. Nature 361, 258–260.

Beaulieu, C., 2011. What makes diffusion anisotropic in the nervous system? In: Jones, D.

(Ed.), Diffusion MRI: Theory, Methods, and Applications. Oxford University Press,

New York, pp. 92–109.

Bhattacharya, J., Petsche, H., 2005.Drawing on mind's canvas:differences in cortical inte-

gration patterns between artists and non-artists. Hum. Brain Mapp. 26, 1–14. http://

dx.doi.org/10.1002/hbm.20104.

Buckner, R.L., 2013. The cerebellum and cognitive function: 25 years of insightfrom anat-

omy and neuroimaging. Neuron 80, 807–815. http://dx.doi.org/10.1016/j.neuron.

2013.10.044.

Buckner, R.L., Krienen, F.M., Castellanos, A., Diaz, J.C., Yeo, B.T.T., 2011. Theorganization of

the human cerebellum estimated by intrinsic functional connectivity. J. Neurophysiol.

106, 2322–2345. http://dx.doi.org/10.1152/jn.00339.2011.

Chamberlain, R., McManus, I.C., Brunswick, N., Rankin, Q., Riley, H., Kanai, R., 2014. Drawing

on the right side of the brain: a voxel-based morphometry analysis of observational

drawing. NeuroImage 96, 167–173.

Cohen, D.J., 2005. Look little, look often: the inﬂuence of gaze frequency on drawing accu-

racy. Percept. Psychophys. 67, 997–1009.

Dale, A.M., Fischl, B., Sereno, M.I., 1999. Cortical surface-based analysis: I. segmentation

and surface reconstruction. NeuroImage 9, 179–194.

Demerens, C., Stankoff, B., Logak, M., Anglade, P., Allinquant, B., Couraud, F., Zalc, B.,

Lubetzki, C., 1996. Induction of myelination in the central nervous system by electri-

cal activity. Proc. Natl. Acad. Sci. 93, 9887–9892.

Dietrich, A., Kanso, R., 2010. A review of EEG, ERP, and neuroimaging studies of creativity

and insight. Psychol. Bull. 136, 822–848. http://dx.doi.org/10.1037/a0019749.

Ditye, T., Kanai, R., Bahrami, B., Muggleton, N.G., Rees, G., Walsh, V., 2013. Rapid changes

in brain structure predictimprovements induced by perceptuallearning. NeuroImage

81, 205–212. http://dx.doi.org/10.1016/j.neuroimage.2013.05.058.

Draganski, B., Gaser, C., Busch, V., Schuierer, G., Bogdahn, U., May, A., 2004. Changes in

grey matter induced by training. Nature 427, 311–312.

Driemeyer, J., Boyke, J., Gaser, C., Büchel, C., May, A., 2008. Changes in gray matter induced by

learning—revisited. PLoS One 3, e2669. http://dx.doi.org/10.1371/journal.pone.0002669.

Floyer-Lea, A., Matthews, P.M., 2004. Changing brain networks for visuomotor control

with increased movement automaticity. J. Neurophysiol. 92, 2405–2412. http://dx.

doi.org/10.1152/jn.01092.2003.

Frankenstein, A., 1953.After the Hunt: William Harnett and other American Still LifePain-

ters 1870–1900. University of California Press, Berkeley.

Frick, A., Möhring, W., Newcombe, N.S., 2014. Development of mental transformation

abilitie s. Trends Cog n. Sci. 1–7http://dx.doi.org/10.1016/j.tics.2014.05.011.

Fry, R.E., 1920. Vision and Design. Chatto & Windus, London.

Glazek, K., 2012. Visual and motor processing in visual artists: implications for cognitive and

neuralmechanisms.Psychol.Aesthet.Creat.Arts6,155–167. http://dx.doi.org/10.1037/

a0025184.

Gombrich, E.H., 1960. Art and Illusion: A Study in the Psychology of Pictorial Representa-

tion. Princeton University Press, Princeton.

Goodale, M.A., Milner, A.D., 1992. Separate visual pathways for perception and action.

Trends Neurosci. 15, 20–25.

Graham, D.J., Meng, M., 2011. Lightness constancy in artists. J Vis. 11, 371.

Guilford, J.P., 1967. The Nature of Human Intelligence. McGraw-Hill, New York.

Hanke, M., Halchenko, Y.O., Sederberg, P.B., Hanson, S.J., Haxby, J.V., Pollmann, S., 2009.

PyMVPA: a python tool box for multivariate pattern analysis of fMRI data.

Neuroinformatics 7, 37–53. http://dx.doi.org/10.1007/s12021-008-9041-y.

Hee Kim, K., 2006. Is creativity unidimensional or multidimensional? Analyses of the Tor-

rance tests of creative thinking. Creat. Res. J. 18, 251–259. http://dx.doi.org/10.1207/

s15326934crj1803_2.

Ishibashi, T., Dakin, K.A., Stevens, B., Lee, P.R., Kozlov, S.V., Stewart, C.L., Fields, R.D., 2006.

Astrocytes promote myelinatio n in response to electrical impulses. Neuron 49,

823–832. http://dx.doi.org/10.1016/j.neuron.2006.02.006.

Jung, R.E., Grazioplene, R., Caprihan, A., Chavez, R.S., Haier, R.J., 2010. White matter integ-

rity, creativity, and psychopathology: disentangling constructs with diffusion tensor

imaging. PLoS One 5, e9818. http://dx.doi.org/10.1371/journal.pone.0009818.

Kowatari, Y., Lee, S.H., Yamamura, H., Nagamori, Y., Levy, P., Yamane, S., Yamamoto, M.,

2009. Neural netw orks involved in artistic creativity. Hum. Brain Ma pp. 30,

1678–1690. http://dx.doi.org/10.1002/hbm.20633.

Kozbelt, A., 2001. Artists as experts in visual cognition. Vis. Cogn. 8, 705–723. http://dx.

doi.org/10.1080/13506280042000090.

Kozbelt,A., Seeley, W.P., 2007. Integrating art historical, psychological,and neuroscientiﬁc

explanations of artists' ad vantages in drawing and perception. Psychol. Aesthet.

Creat. Arts 1, 80–90. http://dx.doi.org/10.1037/1931-3896.1.2.80.

Limb, C.J., Braun, A.R., 2008. Neural substrates of spontaneous musical performance: an

FMRI study of jazz improvisation. PLoS One 3, e1679. http://dx.doi.org/10.1371/

journal.pone.0001679.

Lövdén, M., Bodammer, N.C., Kühn, S., Kaufmann, J., Schütze, H., Tempelmann, C., Heinze,

H.-J., Düzel, E., Schmiedek, F., Lindenberger, U., 2010. Experience-dependentplasticity

of white-matter m icrostructure extends into old age. Neuropsychologia 48,

3878–3883. http://dx.doi.org/10.1016/j.neuropsychologia.2010.08.026.

Maquet, P., Schwartz, S., 2003. Sleep-related consolidation of a visuomotor skill: brain mech-

anisms as assessed by functional magnetic resonance imaging. J. Neurosci. 23,

1432–1440.

May, A., 2011. Experience-dependent structural plasticity in the adult human brain.

Trends Cogn. Sci. 15, 475–482. http://dx.doi.org/10.1016/j.tics.2011.08.002.

Miall, R., Reckess, G., Imamizu, H., 2001. The cerebellum coordinates eye and hand track-

ing movements. Nat. Neurosci. 4, 638–644.

Milbrandt, M., Milbrandt, L., 2011. Creativity: what are we talking about? Art Educ. 8–14.

Miller, E.K., Freedman, D.J., Wallis, J.D., 2002. The prefrontal cortex: categories, concepts

and cognition. Philos. Trans. R. Soc. Lond. B Biol. Sci. 357, 1123–1136. http://dx.doi.

org/10.1098/rstb.2002.1099.

Mishkin, M., Ungerleider, L.G., 1982. Contribution of striate inputs to the visuospatial

functions of parieto-preoccipital cortex in monkeys. Behav. Brain Res. 6, 57–77.

Müller-Lyer, F.C., 1889. Optische urteilstäuschungen. Arch. Anat. Physiol. Physiol. Abt. 2,

263–270.

Norman, K.A., Polyn,S.M., Detre, G.J., Haxby, J.V., 2006. Beyondmind-reading: multi-voxel

pattern analysis of fMRI data. Trends Cogn. Sci. 10, 424–430. http://dx.doi.org/10.

1016/j.tics.2006.07.005.

Ostrofsky, J., Kozbelt, A., Seidel, A., 2012. Perceptual constancies and visual selection as

predictors of realistic drawing skill. Psychol. Aesthet. Creat. Arts 6, 124–136. http://

dx.doi.org/10.1037/a0026384.

Perdreau, F., Cavanagh, P., 2011. Do artists see their retinas? Front. Hum. Neurosci. 5,

1–10. http://dx.doi.org/10.3389/fnhum.2011.00171.

Perdreau, F., Cavanagh, P., 2013. The artist's advantage:better integration of object infor-

mation across eye movements. Iperception 4, 380–395. http://dx.doi.org/10.1068/

i0574.

450 A. Schlegel et al. / NeuroImage 105 (2015) 440–451

Perdreau, F., Cavanagh, P., 2014. Drawing skill is related to the efﬁciency of encoding ob-

ject structure. Iperception 5, 101–119. http://dx.doi.org/10.1068/i0635.

Perna, A., Tosetti, M., Montanaro, D., Morrone, M.C., 2005. Neuronal mechanisms for illu-

sory brightness perception in humans. Neuron 47, 645–651. http://dx.doi.org/10.

1016/j.neuron.2005.07.012.

Plewan, T., Weidner, R., Eickhoff, S.B., Fink, G.R., 2012. Ventral and dorsal stream interac-

tions during the perception of the Müller–Lyer illusion: evidence derived from fMRI

and dynamic causal modeling. J. Cogn. Neurosci. 24, 2015–2029. http://dx.doi.org/

10.1162/jocn_a_00258.

Rougier, N.P., Noelle, D.C., Braver, T.S., Cohen, J.D., O'Reilly, R.C., 2005. Prefrontal cortex

and ﬂexible cognitive control: rules without symbols. Proc. Natl. Acad. Sci. 102,

7338–7343. http://dx.doi.org/10.1073/pnas.0502455102.

Ruskin, J., 1857. The Elements of Drawing. Smith, Elder, & Company, London.

Schlegel, A., Rudelson, J.J., Tse, P.U., 2012.White matter structure changesas adults learn a

second language. J. Cogn.Neurosci. 24, 1664–1670. http://dx.doi.org/10.1162/jocn_a_

00240.

Scholz, J., Klein, M.C., Behrens, T.E.J., Johansen-Berg,H., 2009. Training induces changes in

white-matter architecture. Nat. Neurosci. 12, 1370–1371. http://dx.doi.org/10.1038/

nn.2412.

Shiff, R.,2004. Barnett Newman: A CatalogueRaisonne. Yale University Press, New Haven.

Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E.J., Johansen-Berg,

H., Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., Niazy, R.K., Saunders, J.,

Vickers, J., Zhang, Y., De Stefano, N., Brady, J.M., Matthews, P.M., 2004. Advances in

functional and structural MR image analysis and implementation as FSL. NeuroImage

23 (Suppl. 1), S208–S219. http://dx.doi.org/10.1016/j.neuroimage.2004.07.051.

Smith, S.M., Nichols, T.E., 2009. Threshold-free cluster enhancement: addressing prob-

lems of smoothing, threshold dependence and localisation in cluster inference.

NeuroImage 44, 83–98. http://dx.doi.org/10.1016/j.neuroimage.2008.03.061.

Solso, R.L.,2001. Brain activities in a skilled versus a novice artist: an fMRI study.Leonardo

34, 31–34. http://dx.doi.org/10.1162/002409401300052479.

Stiles, K., Selz, P.H.,2012. Theories and Documents of Contemporary Art: A Sourcebook of

Artists' Writings, 2nd ed. University of California Press, Berkeley.

Tanji, J., Hoshi, E., 2001. Behavioral planning in the prefrontal cortex. Curr. Opin.

Neurobiol. 164–170.

Taubert,M.,Draganski,B.,Anwander,A.,Müller,K.,Horstmann,A.,Villringer,A.,Ragert,P.,

2010. Dynamic properties of human brain structure: learning-related changes in cortical

areas and associated ﬁber connections. J. Neurosci. 30, 11670–11677. http://dx.doi.org/

10.1523/JNEUROSCI. 2567-10.2010.

Taylor, I.A., 1976. Psychological sources of creativity. J. Creat. Behav. 10, 193–202. http://

dx.doi.org/10.1002/j.2162-6057.1976.tb01024.x.

Thouless, R.H., 1932. Individual differences in phenomenal regression. Br. J. Psychol. Gen.

Sect. 22, 216–241.

Todorović, D., 1987. The Craik–O'Brien–Cornsweet effect: new varieties and theirtheoret-

ical implications. Percept. Psychophys. 42, 545–560.

Todorović,D.,2002.Constancies and illusions in visual perception. Psihologija 35,

125–207.

Torrance, E.P., 1969. Torrance Tests of Creative Thinking. Personal Press Incorporated.

451A. Schlegel et al. / NeuroImage 105 (2015) 440–451

The role of art knowledge training on aesthetic judgements and executive functions

Article

Full-text available

Feb 2025

The study of how we develop art knowledge can provide valuable insights into the underlying cognitive systems that support expertise and knowledge transfer to new contexts. An important and largely unanswered question is whether art knowledge training impacts subsequent judgements of artworks and executive functions. Across three pre-registered experiments (N > 630 total), which used a training intervention and Bayesian regression modelling, we explore whether art knowledge training impacts subsequent judgements of artworks and executive functions. Experiments 1 and 2 revealed an effect of art training on aesthetic judgements for trained but not untrained artworks. These training effects were generalized to unseen artworks produced by the same artist (Experiment 1) or another artist with a similar style (Experiment 2), but not to different art styles. Experiment 2 also showed that with larger training ‘doses’ (>16 minutes), the generalization effects are stronger. Experiment 3 showed invariance of the attentional network to art training versus non-art training, suggesting similar sensitivity of executive functions to different types of training. This work shines new light on the cognitive systems that support learning and generalization of learning to new contexts. Likewise, from an applied perspective, it emphasizes that learning and generalization can occur rapidly with a relatively short (approx. 16 minutes) training video.

The Temporal Architecture of Aesthetic Experience: Integrating Default Mode Network Dynamics with Learning and Memory Consolidation

Article

Full-text available

Mar 2024

Ramon Kailani

The neural underpinnings of aesthetic experience have emerged as a crucial area of investigation in cognitive neuroscience, yet the temporal dynamics of how these experiences unfold and integrate with learning and memory systems remain poorly understood. This review synthesizes current research in neuroaesthetics, with a particular focus on the role of the Default Mode Network (DMN) in aesthetic appreciation and its temporal evolution. Through analysis of neuroimaging studies and behavioral data, I propose a novel framework for understanding how aesthetic experiences are processed and consolidated over time. Recent findings indicate that the DMN exhibits distinct temporal patterns during aesthetic appreciation, with initial activation followed by dynamic modulation that varies with stimulus appreciation (Belfi et al., 2019). I present evidence that this temporal sequence may be crucial for memory consolidation and learning processes, suggesting a more complex role for aesthetic experience in cognitive function than previously recognized. This paper integrates findings from multiple domains to propose a comprehensive model of how aesthetic experiences are processed, consolidated, and integrated into long-term memory structures, with implications for both theoretical understanding and practical applications in education and therapy.

Art therapy and neuroscience: evidence, limits, and myths

Article

Full-text available

Oct 2024

Christianne E Strang

The evidence base for the effectiveness of art therapy continues to grow, even as a mechanistic understanding of how art therapy works remains limited. One promising avenue for increasing our understanding of how and why art therapy works is through the lens of neuroscience. A neuroscience-based approach to art therapy provides opportunities for improving understanding of the neural processes that underlie the complex interaction between perception, cognition, emotion and behavior that play out in the art therapy process. Understanding how therapeutic change occurs can result in improved treatment and better outcomes for clients. However, it can be tricky to connect art therapy and psychological theory directly to neural responses. The purposes of this perspective are to provide an overview of the current evidence and limits of neurobiological concepts of neuroplasticity, mirror systems, and interoception as applied to art therapy practice, and to provide updated information about outdated concepts that are still actively used in clinical practice. Critical analysis and understanding of the current scientific knowledge base can then be used to guide art therapy practice and support the development of hypothesis-based research to determine the primary mechanisms that drive the observed effects of art therapy interventions.

The Impact of Visual Art-Based Educational Interventions on Creativity: Α Meta-Analysis

Article

Feb 2025
CREATIVITY RES J

A Study of EEG Characteristics of Aesthetic Perception and Assessment of Painting Works during Museum Visits: Neuroaesthetic Study

Article

Feb 2025
Hum Physiol

Psychiatric Disorders and Cognitive Fluctuations in Parkinson’s Disease: Changing Approaches in the First Decades of the 21st Century

Article

Full-text available

Dec 2024
BSRCCS

Parkinson’s Disease (PD) is a multifaceted neurodegenerative disorder characterized, in addition to the well-recognized motor disturbances, by a complex interplay between cognitive and psychiatric manifestations. We dissect the complex landscape of PD-related psychiatric symptoms, taking into account the impact of functional neurological disorders, somatic delusions, impulse control disorders, and conditions within the bipolar spectrum. The newer entities of somatoform and functional neurological disorders, as well as preexisting bipolar spectrum disorders, are analyzed in detail. Moreover, we emphasize the need for a holistic understanding of PD, wherein the cognitive and psychiatric dimensions are valued alongside motor symptoms. Such an approach aims to facilitate early detection and personalized interventions, and enhance the overall quality of life for individuals suffering from this neurodegenerative disorder.

EEG-Characteristics of Aesthetic Perception and Evaluation of Artworks During a Museum Visit: А Neuroaesthetic Study

Article

Oct 2024

Twenty-eight people (30–70 years old, 12 men, 16 women, artists and non-artists) participated in a comparative neuroaesthetic study in the conditions of real-life visit to M. Vrubel mono-exhibition (Russian Museum, St. Petersburg). During the visit to the exhibition (usually lasted for around 60 min), EEG of participants was recorded. The subjects were looking at the paintings for 30 seconds to 3 min and evaluated the subjective aesthetic “attractiveness” of the paintings by a series of button presses (from 1 to 10). Were analyzed EEG spectral power in the α1 (8–10 Hz)-, α2 (10–13 Hz)-, β1 (13–18 Hz)-, β2 (13–30 Hz)-frequency bands during viewing the most famous Vrubel paintings (“Bogatyr”, “Swan Princess”, “Swan”, “Sitting Demon”, “Flying Demon”, “Pan”, etc.) and event-related EEG synchronization/desynchronization in relation to the subjective emotional and aesthetic evaluation of these paintings. Professional artists showed lower spectral power values in α1 (leads F3, C3, T4, Pz) and α2 (F3, Fz, F4, C3, Cz, C4, P3, Pz, P4)-frequency bands in frontal, central, and parietal regions. The artists were also found to have lower power values in β1,2-frequency bands in frontal (F3, F4, C3) regions and higher power values in occipital (O1, O2 – β1, β2) and posterior temporal (β2) cortical regions compared to the group of subjects with no special artistic education. Moreover, artists decision-making about the high emotional-aesthetic attractiveness of paintings was accompanied by an increase in event-related EEG synchronization for 11.5–27 Hz in frontal and central cortical areas over 580–360 ms before giving the response, compared to non-artists, whereas low emotional-aesthetic evaluation was characterized by 9–27 Hz EEG desynchronization, which started 60 ms before the giving the response and lasted up to 440 ms after it, in the posterior temporal and parietal regions. The differences in frontal cortical areas may indicate a higher engagement of the reward system during the perception of aesthetically pleasing paintings, and the differences in parietal and posterior temporal areas may indicate a continuing visual synthesis (more sustained visual attention) during the perception of subjectively less attractive paintings in artists compared to non-artists.

The arts and hemispheric specialization

Chapter

Jan 2025

Dahlia W. Zaidel

Acknowledging Creative Thinking Skills: Educating for a Creative Future

Article

Full-text available

Jan 2024

Elaine Holt

A shared code for perceiving and imagining objects in human ventral temporal cortex

Preprint

Oct 2024

Mental imagery is a remarkable phenomenon that allows us to remember previous experiences and imagine new ones. Animal studies have yielded rich insight into mechanisms for visual perception, but the neural mechanisms for visual imagery remain poorly understood. Here, we first determined that ∼80% of visually responsive single neurons in human ventral temporal cortex (VTC) used a distributed axis code to represent objects. We then used that code to reconstruct objects and generate maximally effective synthetic stimuli. Finally, we recorded responses from the same neural population while subjects imagined specific objects and found that ∼40% of axis-tuned VTC neurons recapitulated the visual code. Our findings reveal that visual imagery is supported by reactivation of the same neurons involved in perception, providing single neuron evidence for existence of a generative model in human VTC. One Sentence Summary Single neurons in human temporal cortex use feature axes to encode objects, and imagery reactivates this code.

Neuroplasticity: changes in grey matter induced by training

Article

Full-text available

Jan 2004

Neuroplasticity: changes in grey matter induced by training

Article

Full-text available

Feb 2004

Does the structure of an adult human brain alter in response to environmental demands? Here we use whole-brain magnetic-resonance imaging to visualize learning-induced plasticity in the brains of volunteers who have learned to juggle. We find that these individuals show a transient and selective structural change in brain areas that are associated with the processing and storage of complex visual motion. This discovery of a stimulus-dependent alteration in the brain's macroscopic structure contradicts the traditionally held view that cortical plasticity is associated with functional rather than anatomical changes.

The nature of human intelligence

Book

Jan 1967

J P Guilford

What Makes Diffusion Anisotropic in the Nervous System?

Article

Nov 2010

Christian Beaulieu

Water molecules diffuse more rapidly along nerve or white matter tracts than across them, resulting in anisotropic diffusion that can be quantified by diffusion tensor imaging (DTI). In order better to interpret DTI results, an understanding of the potential relationships between diffusion measurements and the underlying tissue microstructure is required, as well as recognition of the limitations in this regard. This chapter presents many of the basic diffusion MRI studies that have attempted to sort out the potential contributions from magnetic susceptibility, the cytoskeleton, axonal membranes, and myelin to the degree of anisotropic water diffusion.

Art and Illusion: A Study in the Psychology of Pictorial Representation

Article

Sep 1960
J Philos

Individual differences in phenomenal regression

Article

Jan 1932

R.H. Thouless

Optische Urteilstäuschungen

Article

Jan 1889

F C Müller-Lyer

Lightness constancy in visual artists

Article

Sep 2011
J VISION

Research to date has produced widely divergent findings in terms of the degree of lightness constancy shown by individuals, and there are hosts of competing models of lightness processing in the brain. We argue here that visual artists–especially painters of representational works–are a useful cohort for investigations of lightness perception because the task of creating representational paintings would appear to require artists to judge natural scene luminance (what artists call “value”) accurately in the face of visual system efforts to normalize perceived intensity. Here we investigate the ability to judge illusory luminance differences in a Craik-O'Brien-Cornsweet stimulus in artists with varying expertise. In a 2AFC staircase design, professional realist painters (N = 2) show lower overall PSE (point of subjective equality) with respect to real luminance differences, compared to student artists (N = 3) and non-artists (N = 9) considered together. For artists vs. non-artists, this result holds at 2 of 3 levels of illusory contrast (i.e., the contrast of the luminance ramp in the center of the illusory stimulus). In addition, there is an inverse correlation between the subjects' reported number of hours spent painting or drawing per week and PSE for the same two illusory contrasts (R² ≥ .18). Artistic training regimes–and perhaps innate endowments–that lead to the putative ability to judge luminance accurately may be seen as analogous to those that allow superior pitch judgment (e.g., absolute or relative pitch) in musicians, and therefore we provisionally term superior luminance judgment as absolute (or relative) value. Artists with this ability are of interest to vision science, much as musicians with superior pitch perception have long been studied by audition researchers to elucidate neural processing of acoustic tone. Tests aimed at determining the factors involved in artists' ability to judge luminance and tests quantifying luminance judgment in naturalistic conditions are also described.

The Elements of Drawing

Article

Oct 1974

After the Hunt: William Harnett and Other American Still Life Painters, 1870-1900

Article

Sep 1978
Coll Art J

The artist emerges: Visual art learning alters neural structure and function

Figures

Recommended publications

Increased cerebellar activation after repetitive transcranial magnetic stimulation over the primary...

Current Progress in CNS Imaging of Myotonic Dystrophy

GAP43 in the cat visual cortex during postnatal development

Effects of Shortened Mandibular Dental Arch on Human Brain Activity during Chewing : An fMRI Study