Neuroimaging the Effect of Popularity on Music Preference in Adolescents

1
Neuroimaging the Effect of Popularity on Music
Preference in Adolescents
Gregory S. Berns with Monica Capra, Sara Moore,
and Charles Noussair Dept. of Psychiatry
Behavioral Sciences, Emory University School of
Medicine Economics Department, Emory
University Goizueta Business School Supported by
grants from the National Institute on Drug Abuse
(NIDA)
2
General Area of Interest

• What are the mechanisms by which social
  information affects consumption decisions?
• Two possible mechanisms
• Informational / normative (Action)
• Preferential (Constructed preference hypothesis
  CPH)
• We use brain imaging technology to discriminate
  between these

3
Why Brain Imaging?

• Difficult to isolate mechanisms with traditional
  laboratory techniques
• We can observe whether conformity results from
• A change in intrinsic valuation CPH
• The desire to experience utility from matching
  others
• The desire to avoid disutility from being
  different
• All or a combination of motives

4
Specific Research Strategy

• Effects of the popularity rating of a song on an
  adolescents own rating
• Music consumption good
• Easily delivered and consumed while the subject
  is being scanned
• Previous research done on the process of music
  stimulus (Koelsch, et al , 2005 Grahn Brett,
  2007)

5
Specific Research Strategy

• Adolescents 12-17 subject pool
• Cohort believed to be highly responsive to social
  influence (Steinberg Monahan, 2007)
• Typical consumers of music, responsible for gt30
  of all single album consumption in the US and
  perhaps greater when on-line purchases are
  included

6
Specific Research Strategy

• Rating
• Familiar method of information transmission for
  our participants
• Simple numerical measure of conformity
• Rating reflected WTP for a song as the songs with
  highest ratings were given to the subject in a CD
  at the end of the experiment

7
Experimental design and procedures

• 27 participants ages 12 17
• Average age 14.6
• 14 were female 13 were male.
• Each individual participated at a separate time.
• Sessions conducted at the Emory University
  Hospital, located in Atlanta, Georgia, USA.
• Sessions lasted, on average, two hours.
• Each individual received a payment of US40 for
  his participation (and a music CD).

8
Songs and popularity

• 20 songs downloaded between Oct 23 Nov 8, 2006
  from MySpace.com of unsigned artists in each
  genre.
• Rock, Country Alternative/Emo/Indie
  Hip-Hop/Rap Jazz/Blues and Metal
• Popularity of songs determined by number of
  downloads (quintiles within each category).
• Popularity scores were 1 5 star rating
• We extracted a 15-second clip from each song,
  which we used for the experiment

9
Timing of decisions

• Before entering the scanner, subjects chose their
  three most favorite genres.
• While in the scanner, each participant observed a
  computer screen and used a handheld device to
  submit decisions.
• Each trial consisted of listening to a song clip,
  rating it, and listening it again and rating it
  for a second time
• There were 60 trials (40 with popularity info
  between first and second listens, and 20 without
  popularity info)

10
Trial Design
11
Behavioral Results (All Trials)
12
Behavioral Results (Changed Trials Only)
13
Model of Music Preference
Change in rating preference
Difference between popular rating and initial
preference rating
14
Distribution of Subject ?s
0
0.5
Mean ? 0.15 .02
15
Results Brain activation regions (First Listen)
16
Neural Correlations with Initial Preference
Rating(Listen1 x Rating1)
17
Main Effect of Popularity(Listen2pop-
Listen2nopop)
18
Interaction of Popularity with Subject Behavioral
Parameter (Listen2pop- Listen2nopop) x ?j
Insula
ACC
19
Summary of results

• Behaviorally, observing popularity ratings
  affects ratings of songs
• We find that music activates a large area of
  networks in the brain
• In contrast, very restricted (reward) areas
  respond to the likeability of the song
• We find that the higher the rating of a
  particular song is, the greater is the activation
  in the caudate nucleus.

20
Summary of results

• When popularity is observed, we do not observe
  any activation in regions traditionally
  associated with value (eg caudate)
• The more others like a song does not suggest the
  adolescent subject will like it more

21
Summary of results

• Individuals who exhibited conformity in behavior
  displayed greater activation to popularity
  information in the anterior insula (relative to
  no popularity information)
• ? Insula activation suggests mismatch anxiety
  which seems to motivate a subset of individuals
  to change their initial ratings
• ? We see this as evidence that conformity is
  being generated by feeling bad when one is
  different

22
What are the mechanisms by which social
information affects ratings?

• At least for music and among adolescent subjects
• Social information affects decisions without
  having an effect on preferences
• Subjects change their song ratings to match
  popularity because they experience mismatch
  anxiety
• Conformists seem motivated by the avoidance of
  disutility and not by the pursuit of utility

23
Policy implications ??

• Peer influence is a physiological experience that
  can be measured
• At least in theory, the effectiveness of
  alternative sources of peer pressure can be
  evaluated before implementing them in the field