Embed Size (px)
Citation preview
Hedonic gLMS: a new scale that permits valid
hedonic comparisons
Linda BartoshukSmell & Taste Center
International Citrus & Beverage Conference (ICBC)September 14-17, 2010
Outline of talk
1. Introduction to supertasters
2. Problem: Conventional scales make invalid comparisons across subjects (i.e., scales cannot “see” supertasters).
3. Solution: It’s not easy, but valid comparisons are possible.
Do we all live in the same taste worlds?
NO
Super taster s experience more intense oral sensations than the rest of us.
PROP DEMONSTRATION
• PROP is a medication used to suppress thyroid function in patient’s with Grave’s disease. Typical dose: 200 mg/day.
• PROP paper: about 1.6 mg of PROP
This tongue has been swabbed with blue food coloring. The structures that house taste buds (fungiform papillae do not
stain as well as the rest of the tongue so they appear lighter. Keep a mental picture of this tongue.
This is the tongue of a super taster : a person born with an unusually large number of
fungiform papillae.
• We count the papillae in a circular template 6 mm in diameter just to the side of the midline.
Supertasters can have up to about 60 fungiform papillae in the template area.
Most individuals have far fewer. This individual has 16 fungiform papillae in the template area. Values as low as 5 are normal.
Mag
nitu
de E
stim
ate
Tas
te I
nten
sity
(n
orm
aliz
ed to
tone
s)
Magnitude Estimate Bitterness of .0032 M PROP
(normalized to tones)
0 10 20 30 40
Citric Acid (r =.22, p<.05)
Sucrose r =.46, p<.0001
0
10
20
30
40
50 QHCl r =.62, p<.0001
NaCl (r =.49, p<.001)
10
20
30
40
50
00 10 20 30 40
TASTE: Supertasters have the most fungiform papillae (the structures that house tastebuds) so they have the most tastebuds and thus perceive the most intense tastes.
ORAL BURN: Tastebuds are surrounded by fibers thought to mediate oral burn; thus supertasters perceive greater burn from oral irritants
like chilis.
30020010000
r = .56, p < .0001
Ora
l bur
n
Bitterness of .0032 M PROP
0
100
200
300
400
500
600100 ppm capsaicin
Snyder Senior Essay, Yale University, 1996
Jalapeno peppers contain about 100 ppm capsaicin.
Duffy, Bartoshuk, Lucchina, Snyder &, Tym, 1996.
ORAL TOUCH: Fungiform papillae are innervated by fibers mediating touch; thus supertasters perceive more intense touch sensations (e.g.,
creaminess) from fats C
ream
ines
s
heavy cream r = .44, p < .001
100
9080
70
6050
40
3020
10
0NT MT ST
In addition, there is a connection between supertasting and flavor
(flavor is taste plus RETRONASAL OLFACTION)
Food and Beverage Industry
• Has long known that adding a taste to a beverage (e.g., adding sugar) will intensify the perception of the flavor (e.g., orange) of the beverage.
• Similarly, experiencing greater taste (as do super taster s) will intensify the perception of flavor.
Nontasters
Supertasters
RetronasalOrthonasal
Orthonasal
Retronasal
• Supertasters live in a neon food world.
• Those with the fewest taste buds live in a pastel food world.
I’ve compared supertasters to others with regard to how they perceive the food world. But we cannot share each other’s experiences directly.
How did I make comparisons between supertasters and others?
I did not use the classic Natick 9-point scales (or any similar
scales like VAS).
Natick 9-point scales
1
2
3
4
5
6
7
8
9extremely strong
extremely weak
Sensory Hedonic
1
2
3
4
5
6
7
8
9like extremely
dislike extremely
neutral
The problem.
• What does “extremely strong” mean?
• What does “like extremely” mean?
• Let me begin with the sensory descriptor: “extremely strong.”
• In ordinary conversation, we use intensity descriptors to communicate with each other and compare experiences. “That lemonade tastes extremely strong to me. Does it taste extremely strong to you?”
• Labeled scales use those same descriptors to denote perceived intensities.
• We seem to be communicating, but are we?
• No! “extremely strong ” may denote different perceived taste intensities to each of us.
• How do our subjects interpret “extremely strong” on the 9-point scale?
• Do they tend to interpret “extremely strong“ as referring only to foods when they are in a food experiment?
• Let’s ask them.
We asked our subjects to rate their favorite and least favorite foods.
• The median for favorite food was “9.”
• The median for least favorite food was “1.”
• This tells us that the Natick 9-point scale was interpreted as referring to food preferences only.
12345
6789like extremely
dislike extremely
neutral
favoritefood
least favoritefood
But consider how easily we switch contexts for ratings.
Example: elastic intensity scale
• A woman who has just experienced childbirth may describe her pain as “very strong.”
• Given a cup of tea, she may also describe the flavor of the tea as “very strong.”
Elastic Intensity Scale
We understand that she does not mean to suggest that her pain and the intensity of the tea flavor are the same. She means her pain was “very strong” in the context of all pains she has experienced and the tea flavor was “very strong” in the context of all teas she has sampled.
Comparison errors caused by elasticity of our intensity scale
• What happens when we treat our elastic scale as if it is absolute?
• We first encountered this kind of error in our taste studies involving supertasters.
REVERAL ARTIFACT (shown by r ed dashed lines)
Very Strong
Taste
REALITY
Others ST
Quinine
SucroseNaCl
Others ST
INCORRECT ASSUMPTION that " Very Strong Taste" is the same
absolute intensity for nontasters (NT) and supertasters (ST)
QuinineSucroseNaCl
Very Strong
Taste
Very Strong
Taste
We are not the first to note this error.• Aitken, R. C. B. (1969). Measurement of feelings using visual
analogue scales. Proceedings of the Royal Society of Medicine, 62, 989-993.
• Narens, L., & Luce, R. D. (1983). How we may have been misled into believing in the interpersonal comparability of utility. Theory and Decision, 15, 247-260.
• Biernat, M., & Manis, M. (1994). Shifting standards and stereotype-based judgements. Journal of Personality and Social Psychology, 66, 5-20.
• Birnbaum, M. H. (1999). How to show that 9>221: Collect judgements in a between-subjects design. Psychological Methods, 4, 243-249.
Illustration: PROBLEM and SOLUTION
• Select two groups– people with many fungiform papillae: super tasters– people with few fungiform papillae
• Ask both groups to describe the sweetness of a coke– Both groups rate the sweetness to be about 2/3 of the distance from
no sweet (0) to the strongest sweet they have ever tasted (100%).– Thus both groups seem to be experiencing the same sweetness.
• Now ask both groups to match the sweetness to the loudness of a tone.
Correct conclusion: Super tasters perceive twice as much sweetness as do Others.
Each 10 db doubles loudness
subway
train whistle
telephone dial tone
loud conversation
90
80
100
70
db
Others
Supertasters
Sweetness of a coke
Summary of Magnitude Matching
• Select a standard that is independent of the sensation we want to compare and ask subjects to rate the sensation of interest relative to the standard.
• If the standard is roughly equal, on average, to two groups, then we can make an absolute comparison of the sensation of interest across the two groups.
• This method was first used in taste (1975) and was formalized as “magnitude matching” in 1980 (Marks & Stevens).
Let’s make this user friendly.
• Let’s take the labeled magnitude scales we are used to (e.g., Natick 9-point scale).
• Respace the intensity descriptors so that the scale has ratio properties (i.e., “8” denotes an intensity twice that of “4”).
• This is essentially the LMS (Labeled Magnitude Scale) devised by Barry Green and his colleagues to be used with oral sensations (Green et al, 1993).
• Green and his colleagues empirically re-spaced the intensity descriptors to give this scale ratio properties
• Turns out that we don’t even really need all of those descriptors.
very strong
strong
moderate
weak
strongest imaginable oral sensation
no sensation
LMS
Evolution of the general Labeled Magnitude scale (gLMS)
0
10
20
30
40
50
60
70
80
90
100
very strong
strong
moderate
weakbarely detectable
strongest imaginable sensation of any kind
no sensation
gLMS
barely detectable
very strong
strong
moderate
weak
strongest imaginable oral sensation
no sensation
LMS
no sensation
strongest imaginable sensation of any kind ever experienced
gVAS
where 100 = 0 = no sensation
strongest sensation of any kind ever experienced
Numbers 0 to 100
Bartoshuk, Puentes, Snyder & Sims, unpublished data
very strong
medium
very weak0123456789
Natick 9-point category scale
Citr ic Acid
Density of Fungiform Papillae (# in template)
0 10 30 4020 50
Perc
eive
d Ta
ste
Inte
nsity
Density of Fungiform Papillae (# in template)
Perc
eive
d Ta
ste
Inte
nsity
Citric Acid
0 10 30 4020 50
very strong
strong
moderateweak
barely detectable
strongest imaginable
sensation of any kind
0
gLMS
Density of fungiform papillae gives us a way to check how good our scales are.
Let’s create a hedonic version of the gLMS
Sensory
0
20
40
60
80
100most intense sensation of
any kind ever experienced
no sensation
Hedonic
-100
-80
-60
-40
-20
0
20
40
60
80
100most intense pleasure of
any kind ever experienced
neutral
most intense displeasure of any kind ever experienced
12345
6789like extremely
dislike extremely
neutral
favoritefood
least favoritefood
-100
-80
-60
-40
-20
0
20
40
60
80
100most intense pleasure of
any kind ever experienced
neutral
most intense displeasure of any kind ever experienced
favorite food
least favoritefood
Natick 9-point scale
Hedonic gLMS
Examples of what the hedonic gLMS can do
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100
favorite food
least favorite food
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8 9
r = .007 p = .92
r = .10 p = .16
r = .40 p < .0001
r = .35 p < .0001
hedonic gLMS hedonic 9-pt
SupertastersOthers SupertastersOthers
Note that the hedonic gLMS reveals that supertasters have more extreme food likes and dislikes.
Hed
onic
gLM
S
-100
-75
-50
-25
0
25
50
75
100
r = .16 p = .02
neutral
Dis
like
L
ike
Dis
like
L
ike
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100
r = .24 p <.0006
SupertastersOthers
neutral
• The gLMS shows that supertasters like orange juice and dislike grapefruit juice more than do others.
• The 9-point hedonic scale cannot reveal this.
These data were collected at lectures so we could
accumulate large samples.
When to use the hedonic gLMS
• When you compare samples, this is a within subject comparison and virtually any scales can do this.
• However, when you need to compare groups they may be quite different with regard to food perception or liking, the older scales may provide erroneous comparisons.
