Cosmetic Science Technology 2013
Establishing Hair Care Sensory Testing Capabilities
Charlotte Stricane, Dr. Robert M. Sayer, Croda, Ditton, United Kingdom
Abstract
Sensory evaluation is a scientific discipline used to investigate human response to a
stimulus involving the five senses. Sensory properties can be assessed using a wide range
of test methods. Today, the sensorial aspect of a product is directly influencing consumer’s
choices and preferences.
Croda is a global leader in speciality chemicals, sold into a wide range of markets from
Consumer Care and Personal Care to Industrial Specialities. Within the Sun Care &
Biotechnology business, the company has recently developed an in-house trained sensory
panel to meet the growing demand for consumer relevant data, in the hair care market. This
article presents a training method to establish a hair care sensory panel and examines a
selection of existing products in the Croda Biopolymer portfolio.
Introduction
Hair is composed of 70% proteins. These proteins can be classed in two groups: Keratins
and Keratin Associated Proteins (KAPs). Croda has more than 40 years’ experience in
manufacturing protein derivatives in order to address the needs of the hair care market.
Sensory evaluation is “a scientific method used to evoke, measure, analyse and interpret
reactions to those characteristics of food and materials as they are perceived by the senses
of sight, smell, taste and hearing” [1].
This definition was written for the food industry, as it was the first to investigate the human
reactions to measured stimuli. It has since been adopted by many other industries such as
cosmetics, paints and automotive.
Subjective studies collect data about a consumer’s preferences while objective studies
establish perception thresholds and quantified responses.
The objective studies require that a group of people are trained to give quantitative answers,
unbiased by their own preferences. Objective sensory includes tests such as triangle, paired
and ranking tests [2]. To evaluate several attributes and products in one study, a profiling
1
Cosmetic Science Technology 2013
test is preferred where each attribute evaluated is marked against a quantitative scale and
against a control.
Hair sensory evaluation is a rising discipline in the Personal Care industry. Companies are
developing in–house methods to collate sensory data that is used to create consumer
relevant marketing packages. These techniques involve a source of hair (ex-vivo tresses or
in-vivo salon testing) and a trained panel to assess the characteristics of the hair and
treatments.
There are different types of sensory panel that can be established, depending on the
question to be addressed, subjective or objective, and the resources available. To produce
a statistically reliable data set the number of panellists required can vary depending on their
level of training. The three main types of panel are;
•
An untrained or consumer panel will require between 50 and 200 panellists.
•
A trained panel is generally composed of 10 to 20 panellists, trained on a regular
basis to rank the product against set benchmarks.
•
An expert panel is a group of two to five people as reliable as machines, trained
every week and running sensory evaluation studies as a full time job.
The present article describes setting up a trained panel composed of 15 people studying the
effects of protein derivatives on the sensorial aspects of human hair.
Protocol development
For a given product, each attribute is marked against a numerical scale from 1 (poor), to 9
(good). Like any new piece of equipment, the panel needs to be calibrated in order to
produce accurate measurements.
The calibration step can be a long process but is necessary to achieve statistical
reproducibility.
During the first session conducted by an external consultant, panel members are introduced
to a list of characteristics or attributes of hair and are asked to give their own definition,
prompted by a sample of high and low benchmark for each attribute. When a consensus is
reached, the definition of the attribute is agreed and the discussion moves on to the next
attribute. Croda’s panel uses 10 attributes on both wet and dry hair:
2
Cosmetic Science Technology 2013
Wet hair attributes:
-
Detangling: 1 = long time to detangle and 9 = no knots to detangle
-
Combing: 1 = very hard to comb and 9 = very easy to comb
-
Smoothness: 1 = rough sensation and 9 = sleek sensation
Dry hair attributes:
-
Visual Clean appearance: 1= fibres stuck together in bundles and 9 = separated
fibres
-
Shine: 1 = dull and 9 = very shiny
-
Clean feel: 1 = heavy residue on the hair/fingers and 9 = no residue
-
Combing: 1 = very hard to comb and 9 = very easy to comb
-
Smoothness: 1 = rough sensation and 9 = sleek sensation
-
Softness: 1 = stiff hair and 9 = supple hair
-
Fly away Control: 1 = important flyaway and 9 = no flyaway
The benchmark testing is repeated as often as possible, using a set protocol and samples
coded with random numbers. The evaluation order is also randomised to remove any source
of bias. The protocol shown in Figure 1 is for one of the attributes and assesses three sets of
hair tresses, each set including a shampooed tress (also called ‘control’) and two other are
treated with different products. This randomised triplicate testing method ensures the panel
generates robust data for statistical interrogation.
Figure 1. Example of an attribute and its associated protocol for benchmark testing
Calibration is achieved when the panel statistically agree on the scores for each benchmark
product and for each of the 10 attributes. This process is monitored using statistics such as
individual mean, bias and standard deviation between repeats.
Monitoring the panel over time
Individual bias is an important parameter in monitoring a panel. For each attribute, it directly
reflects the difference between the panellist result and the panel average. To show
improvement, individual bias should show a convergence to zero. In practice, it has been
established that a bias of +/- 1 is acceptable.
3
Cosmetic Science Technology 2013
Figure 2. Individual bias over time
By plotting Individual bias over time, as shown in Figure 2, the panel leader is able to
visualise each panellist’s performance and compare new panellists to the others. In this
example, the blue panellist shows an improvement over time, converging with the group
(e.g. decreasing bias) whereas the brown panellist can be identified as a regular outlier for
this studied attribute. By monitoring the evolution of bias for the different attributes, the panel
leader is able to correct ‘regular outliers’ by providing more targeted training.
Protocol adjustment resulting from panel monitoring
In order to compare different studies, a control must be included. Croda’s basic shampoo
was chosen for its neutrality on sensorial properties of hair. Figure 3 presents the records of
the control profile over nine consecutive studies. Sensory work is presented in the form of a
spider diagram where attributes are plotted in a circular plot with score values of 1- 9.
4
Cosmetic Science Technology 2013
Figure 3. Spider diagram of control profiles in grey and the average trace in red.
Fig 3 demonstrates panel consistency when assessing the shampoo tresses over a
prolonged period of time using the randomised testing protocols.
Data for nine studies were analysed with a One Way ANOVA test. This test is used to
ensure the profiles are not significantly different from one another within the group.
The attributes showing ‘differences’ can be a function of the hair preparation stage,
individual perception threshold or factors such as a fluctuation of the relative humidity levels
in the testing area. Once identified, these issues are addressed by adapting protocols and
methods. The average profile, in red on Figure 3, can then be used as the reference or
control profile in future studies.
Protocol adjustment resulting from panel reproducibility
By analysing the shampoo profile associated with repeated studies of a quaternised product,
a large variability was found to be caused by the hair condition altering over time.
5
Cosmetic Science Technology 2013
Figure 4. Spider diagram of shampoo profiles from week 1 to week 6
Figure 4 shows wet attributes scoring higher with the number of studies over time. This
behaviour is characteristic of an accumulation of a conditioning substance on the surface of
the hair. The hair baseline shifts as a result of conditioning.
Consequently, the washing protocol during hair preparation was modified to include a
stripping step using a denatured alcohol rinse. Tress management was critical to prevent
over use or over conditioning of the hair.
Protein derivatives
Full profiling data, including the assessment of the control and two different protein actives,
are plotted in spider diagrams and the data treated with ANOVA two-way tests. It allows the
method to assess the ‘equality of the means’ when two factors are influencing the results,
e.g. the products and the assessors. Products (treatments) are compared against each other
to highlight any differences and the assessors (panellists) are compared against each other
to reveal any disagreement or shift in panel calibration. The test is run with MINITAB®
statistical software. In general, for a confidence level of 95%, the P value for treatments must
be under 0.05 for the difference to be significant.
The difference between products or ‘treatments’ is illustrated by Figure 5, where A is the
control, and the different products evaluated are B and C.
6
Cosmetic Science Technology 2013
Figure 5. Visual explanation of ANOVA analysis for ‘treatments’ or products
The distribution of the scores along the scale reflects the intensity of the differences felt by
the panel e.g. in Fig 5 C is very different than A or B. The differences between panellists or
‘assessors’ should be narrow and the P value result above 0.05.
ANOVA studies are followed by Multiple Range Analysis in order to determine the least
significant difference (LSD) between the treatments. To be significantly different from one
another, the difference between treatment means must be equal to or larger than the LSD.
Treatments that do not share a letter are significantly different from one another.
Investigating low molecular weight protein products
Previous mechanical studies showed that small peptide/amino acid products penetrate into
the hair shaft and bind moisture in the cortex [3]. Testing of such ingredients was conducted
with the panel, on three sets of tresses and the products were delivered from 5% active
aqueous solutions.
7
Cosmetic Science Technology 2013
Sample/Product Name (INCI)
Product A
Control
Product B
Crotein WKP (Aqua (and) Hydrolysed Keratin)
Cropeptide W (Aqua (and) Hydrolysed Wheat Protein (and)
Hydrolysed Wheat Starch)
Product C
Figure 6. Spider diagram of short proteins vs control treatment.
In the case of product B, the low molecular weight keratin peptides, and product C, the
medium molecular weight wheat peptides, no significant difference could be observed in this
assessment. The sensory data showed that this type of compound is not detected on the
surface of the hair by a human panel, reinforcing the idea of penetration of low
molecular weight proteins into the hair structure.
Investigating protein derivatives
Cationic compounds are known to target the negative charges along damaged hair fibres
and this improves the condition of the hair. The treatment results in more manageable
healthy looking hair.
Two wheat proteins, derivatised with very different cationic compounds were chosen to
illustrate the influence of the product chemistry on hair sensory and the ability of the panel to
differentiate between the treatments. Both products were delivered from a 5% aqueous
solution and rinsed thoroughly.
8
Cosmetic Science Technology 2013
Sample/Product Name (INCI)
Product A
Control
Product B
Hydrotriticum QM (Aqua (and) Cocodimonium
Hydroxypropyl Hydrolyzed Wheat Protein)
Product C
Voluminis (Aqua (and) Ethyltrimonium Chloride
Methacrylate/Hydrolyzed Wheat Protein Copolymer)
Figure 7. Spider diagram of the cationic ingredients vs control treatment
The inverse solubility of product B, the quaternised hydrolysed protein causes deposition on
the hair. The panel was able to significantly detect it from both visual (Dry CleanAppearance
and Shine) and touch aspects (DryCleanFeel, DrySmoothness and Dry Softness). Product B
still demonstrates some of the key conditioning attributes by significantly improving
TangleRemoval, and FlyAwayControl compared to the control.
On the other hand, Product C was designed to condition the hair by improving
TangleRemoval, WetCombing, WetSmoothness and FlyAwayControl compared to the
control (shampoo) without creating a sensorial difference on dry attributes.
The combination of medium molecular weight proteins and quaternised side chains in
Product C perform as a “light conditioning” ingredient, making it suitable for volumising (or
light) conditioners.
9
Cosmetic Science Technology 2013
Difference
TangleRemoval
Significant
WetCombing
Significant
WetSmoothness
Significant
DryCleanAppearance
Significant
DryShine
Significant
Tukey’s multiple range test
95%
Treatment Mean Group
C
7.3 X
B
4.6
Y
A
3.6
Z
C
6.8 X
B
4.3
Y
A
3.9
Y
C
5.7 X
B
4.4
Y
A
3.8
Y
A
6.6 X
C
6.2 X
B
2.4
Y
A
6.2 X
C
6.0 X
B
4.8
Y
Difference
DryCombing
Significant
DryCleanFeel
Significant
DrySmoothness
Significant
DrySoftness
Significant
FlyAwayControl
Significant
Tukey’s multiple range test
95%
Treatment Mean Group
C
6.4 X
A
5.8 X Y
B
5.5
Y
A
6.3 X
C
5.6 X
B
2.3
Y
A
6.2 X
C
6.1 X
B
4.5
Y
A
5.9 X
C
5.5 X
B
3.9
Y
B
7.2 X
C
5.6
Y
A
3.4
Z
Figure 8. Table summarising statistical treatments of the data
Different hair types
Hair has a different physical form depending on where individuals originate from [4].
Cross sections of Chinese hair are circular ellipses; the hair is thick and appears very shiny
whereas cross sections of Brazilian hair look like flat ellipses, meaning the hair is more likely
to be curly.
Three regional hair types were assessed by the trained sensory panel. Each type of hair
was damaged according to the most common type of damage inherent to the region of the
world it originates from. European hair was bleached, Chinese hair was dyed and Brazilian
hair was relaxed.
The chosen hair types were assessed on an absolute scale and compared over their full
profile.
10
Cosmetic Science Technology 2013
Figure 9. Comparison of the three Hair types
Each hair type offers different sensory properties. For example relaxed Brazilian hair feels
less easy to comb and less smooth in the wet state than Chinese and European hair. The
external elliptical structure of the hair and the nature of the damage created very poor wet
sensory properties. It also highlights that each region of the world requires a customised
regime to help restore their hair’s original sensory properties.
Focusing on biopolymers, the sensory panel has investigated the effects of a quaternised
protein derivative product on damaged European, Chinese and Brazilian hair.
11
Cosmetic Science Technology 2013
Figure 10. Data collected on European bleached, Chinese dyed and Brazilian relaxed hair
after application of the same quaternised protein derivative
The influence of the hair structure and the different type of damage caused to the tresses did
not destabilise the panel’s capacity of assessing the biopolymer’s properties. Compared to
the blank treatment, the biopolymer significantly improves TangleRemoval, Wet Combing,
Wet Smoothness and FlyAwayControl on all of the three hair types, delivering conditioning
benefits from an aqueous solution. The cationic protein derivative’s substantivity to the hair
surface was confirmed by TOF-SIMS imaging and combing studies.
12
Cosmetic Science Technology 2013
Conclusion
Croda has established and trained a sensory panel to work on human hair tresses. The
panel can successfully differentiate 10 hair attributes on three ethnic hair types.
They can accurately assess and quantify biopolymers’ sensory properties on hair delivered
from aqueous solutions. This was achieved by protocol adjustment and thorough monitoring.
The panel demonstrated its strength when testing three different types of hair, as well as
assessing the impact of a cationic protein derivative.
Going forward, the Croda panel will extend their expertise to testing the effects of
biopolymers from shampoo and conditioner systems in order to offer strong and
substantiated sensory claims for the hair care market.
13
Cosmetic Science Technology 2013
References
[1] Anon, Sensory Evaluation Division of the Institute of Food Technologists, (1975)
[2] Principles of Product Evaluation: Objective Sensory Methods, IFSCC Monograph Number
1, Micelle Press
[3] Jones RT, Chahal SP, International Journal of Cosmetic Science, 19, 215-226 (1997)
[4] Robbins CR, Genetic control/Involvement in Hair Fiber traits. In: Chemical and physical
behaviour of human hair, 5th edition. New York: Springer-Verlag; p. 179 (2012)
Authors: Charlotte Stricane and Dr Robert M Sayer
14