lable at ScienceDirect

Food Hydrocolloids 25 (2011) 789e796

Contents lists avai

Food Hydrocolloids

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

Textural perception of liquid emulsions: Role of oil content, oil viscosityand emulsion viscosity

George A. van Aken a,b,*, Monique H. Vingerhoeds a,c, René A. de Wijk a,c

a TI Food and Nutrition, P.O. Box 557, 6700 AN Wageningen, The NetherlandsbNIZO food research, P.O. Box 20, 6710 BA Ede, The Netherlandsc Food and Biobased Research, Wageningen UR, P.O. Box 17, 6700 AA Wageningen, The Netherlands

a r t i c l e i n f o

Article history:Received 9 March 2010Accepted 24 September 2010

Keywords:Food emulsionsSensory perceptionMouthfeelTextureCreaminessThicknessStickinessPolysaccharideThickenerGum arabicTribologyFat reductionFat perception

* Corresponding author. TI Food and Nutrition, P.O.gen, The Netherlands. Tel.: þ31 318 659568.

E-mail address: george.van.aken@nizo.nl (G.A. van

0268-005X/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.foodhyd.2010.09.015

a b s t r a c t

This work describes a study on the in-mouth textural perception of thickened liquid oil-in-wateremulsions. The variables studied are oil content, oil viscosity, and the concentration of polysaccharidethickener. Gum arabic was chosen as the thickener because of the nearly Newtonian behavior of itssolutions and special care was taken to suppress aroma clues. Based on the experimental results andfindings from previous studies, this work shows that the emulsion droplets influence textural sensoryperception of liquid emulsions by three main mechanisms, each of which relate to changes in specificsensory attributes, and none of which were found to be significantly dependent on the viscosity of theoil: 1) by increasing the viscosity, 2) by becoming incorporated in the mucous oral coating, and 3) byspreading oil at the oral surfaces. Based on these results, the possibility for replacement of emulsified fatby a polysaccharide thickener is evaluated.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Triglyceride fat (in liquid form referred to as “oil”) is an impor-tant ingredient in many food systems. It has important functions asa heat transfer medium in the production stage of fried foods, asa barrier protecting against water migration in confectioneryproducts and to prevent crisps from becoming soggy, as a short-ening agent in bakery products, as a structuring agent for examplein butter, chocolate, whipped cream and cake, and as a solvent fordelivery of flavours and vitamins and to soothe bitter off-tastes inmany foods. Moreover, in many food systems emulsified fatcontributes considerably to the in-mouth textural quality, typicallyenhancing smoothness, fullness and creaminess.

Fat is however also the most energy dense of the three mainnutrients (9 kcal/g for fat versus 4 kcal/g for proteins and sugars)and for that reason reducing the fat content forms a main challenge

Box 557, 6700 AN Wagenin-

Aken).

All rights reserved.

for food producers, complying with the general concern about theincreasing number of people suffering from overweight or obesity.In reduced-fat products, the various roles of fat have to be replacedby other ingredients. Focussing on the in-mouth textural sensoryproperties of liquid reduced-fat systems, emulsified fat behaves asa filler, contributing significantly to the viscosity. This effect can bereplaced to a large extent by polysaccharide thickeners. However,the reduced-fat products (for example reduced-fat creams, saucesand ice cream) are commonly described as less creamy and judgedto be less attractive.

Several studies have confirmed that although the perception ofa creamy mouthfeel relies heavily on the rheological properties (forliquid systems the viscosity) of the food product, in addition alsoother aspects are important. For example, the prediction of creamymouthfeel scores from instrumental viscosity improved whentribological data were included demonstrating the importance offriction-reducing properties of fat for creaminess (Kilcast & Clegg,2002; Kokini, 1987; de Wijk & Prinz, 2005). In a study on themouthfeel of emulsions of sunflower oil (0, 5,10 and 40%), stabilizedby a range of emulsifiers (whey protein isolate, sodium caseinate,stearyl lactylate, lecithin, skimmed milk powder, monoglyceride,

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796790

sucrose ester) and for some systems thickened by a range of starchtypes, the ratings of creamy mouthfeel (Mcreamy) were comparedwith the ratings of thick mouthfeel (Mthick) (van Aken, de Hoog,Nixdorf, Zoet, & Vingerhoeds, 2006; van Aken, Vingerhoeds, & deHoog, 2009). An interesting difference between Mthick andMcreamy was found for the systems without starch, for which theviscosity ranged between 1 and 10mPa s.Whereas for systemswith5, 10 and 40% oil the variation in Mthick and Mcreamy was corre-lated and their scoring increased with the viscosity (R2 ¼ 0.67), thesystems with 0% fat showed a similar increase of Mthick with theviscosity, but Mcreamy remained a constant low level independentof the viscosity. This result strongly suggests a specific, fat-depen-dent effect on (creamy) mouthfeel for low-viscosity emulsions, inaddition to a primary role of the viscosity.

Also Akhtar and co-workers (Akhtar, Stenzel, Murray, &Dickinson, 2005) found a specific effect of emulsified oil onmouthfeel, in addition to a major correlation of Mcreamy withMthick for emulsions thickened by polysaccharides. These authorslater reported an important role of the type of thickener (malto-dextrin versus xanthan) (Akhtar, Murray, & Dickinson, 2006) oncreaminess. Even though care had been taken to match theviscosities of the samples at a shear rate of 50 s�1, which wasthought to represent the flow conditions in the mouth duringsensory rating, the systems with maltodextrin were rated creamierthan those with xanthan. As possible explanations of this differ-ence, the authors put forward that xanthan is more shear thinningleading to a lower viscosity at high shear rates, and a possibledifference in the ability of the polysaccharides to coat the oralsurfaces. In this context it should be noted that for polysaccharidesalso other sensory attributes become important, such as theperceived sliminess, slipperiness and stickiness (Nishinari, 2006).

The purpose of the present work was to further investigate theeffect of fat on the in-mouth textural sensory perception of liquidemulsions. To this end we carried out a detailed sensory study oftextural sensory attributes (according to the principles of Quanti-tative Sensory Analysis (QDA) (Stone & Sidel, 2004)) of emulsionsystems which vary in oil content, oil viscosity and thickenercontent. We attempted to separate the role of the emulsionviscosity and the alleged fat-specific effect by matching theviscosities of systems with different oil content. Using oils withdifferent viscosities allowed us to investigate if the alleged fat-specific effect is related to the viscosity of the oil, for examplethrough the formation of a hydrodynamically lubricating oil film onthe surfaces of tongue and palate.

2. Experimental

2.1. Experimental approach

The viscosity of the oil was varied by using medium chaintriglycerides (MCT), olive oil (Olive) and castor oil (Castor). Theviscosity of the emulsionwas varied by adjusting the concentrationof the polysaccharide thickener. In this way the effect of the oilcontent on the viscosity is in principal also minimized, although inreality polysaccharide solutions are also shear thinning, whichintroduces additional sensory attributes. To avoid the effect of shearthinning of the polysaccharide as much as possible, we used gumarabic as the thickener, which because of its highly branched andcompact molecular structure behaves nearly Newtonian in solu-tion, and adjusted the viscosities to the value measured at about50 s�1, which is thought to be typical for thickness perception(Shama & Sherman, 1973a, 1973b; Wood, 1968).

Care was taken to avoid flavor cues from the oil. For thatpurpose, the oils were deflavored by steam distillation. The oilcontent was 20% and the viscosity was adjusted at 4 different levels,

A, B, C and D, targeting viscosities of 2, 133, 335 and 2710 mPa s�1,respectively. For each viscosity level a 2%MCTemulsionwas used asa low-fat reference.

2.2. Materials

Whey protein isolate (WPI), brand name BiPro, had a proteincontent of 97.8% and was obtained from Davisco Foods Interna-tional (US). Distilled water was used for making the solutions.Vanilla flavor (vanilla flavouring powder T03912) containing 25%nature-identical vanillin andmaltodextrin as carrier was a gift fromDanisco Holland BV (The Netherlands). Sugar (CSM, TheNetherlands) and NaCl (NEZO, Akzo Nobel, the Netherlands) werepurchased from a local retailer. MCT (Miglyol 812 N), waspurchased from Internatio NV, Belgium; Olive, Castor (Ricini oleumvirginale) and pharmaceutical grade gum arabic (acaciae gummi(180)) were obtained Spruyt Hillen, IJsselstein, the Netherlands.Batches of 2 kg of the oils were deflavored by steam distillation (3 hat a vacuum of 1e2 kPa) and immediately stored under nitrogen atTNO, Zeist, The Netherlands. All oils were completely Newtonian(variation in viscosity between 10 s�1 and 500 s�1 less than 1%),with viscosities MCT: 30.4 mPa s; Olive: 81.8 mPa s; Castor:984mPa s. The emulsions were prepared under hygienic conditions1 day before panel evaluation and stored at 4e7 �C.

2.3. Emulsion preparation

A stock aqueous solutionwas prepared by dissolving 1wt%WPI ina 10 mM NaCl solution at 20 �C under continuous gentle stirring for1 h. Gum arabic stock solutions were made by dissolving gum arabicpowder inwater under agitation and heating “au bain marie” (about1 h at 80 �C) until all lumps had disappeared. The pH of the aqueousand gum arabic stock solutions were adjusted to pH 6.7 with hydro-chloric acid or sodiumhydroxide, and vanilla flavor and sucrosewereadded to even out small flavor differences between the emulsions.The final composition contained a variable concentration of gumarabic,1.8wt% sucrose, 0.0548wt% NaCl, and 0.05wt% vanillin flavor.

Pre-emulsions containing 40 wt% oil and 1 wt% WPI wereprepared by mixing oil into an aqueous solution containing WPI,sucrose, salt and vanilla flavor using a rotor-stator type mixer at4500 min�1 (Ultra Turrax, IKA, T25-basic, Germany). This pre-emulsion was homogenized using a GEA Nivo Soavi SPA (typeNS1001L2K) high pressure homogenizer operating at 2 subsequentpressure steps; all emulsions received a first homogenization, atpressures adapted for each oil viscosity (25 MPa (MCT), 35 MPa(Olive) and 80 MPa (Castor)), followed by a second homogenizationat 2 MPa to disrupt droplet aggregates. This procedure resulted insimilar average droplet sizes of approximately 1 mm for all emul-sions. The emulsionswere dilutedwith aqueous or gumarabic stocksolutions to the desired oil and gum arabic contents for each tar-geted viscosity. The concentration of gum arabic was calculated foreach targeted viscosity by a correcting a master curve for theviscosity of gum arabic solutions with the Krieger Dougherty rela-tion (Eq. (1) for the volume fraction of the emulsion droplets, in thisway assuming hardesphere interaction between unaggregateddroplets. All emulsions had a neutral pH (6.7), were refrigerateduntil use (within 1 day after preparation) and checked for microbialsafety by Silliker BV (Ede, The Netherlands). All cases showednegligible counts for aerobic mesophilic germination number,Bacillus cereus, Staphylococcus aureus and Listeria monocytogenes.

2.4. Emulsion characterization

Samples were analyzed by light microscopy (Olympus BX 60Microscope equipped with an Olympus DP 70 camera; Olympus

Table 1Description of mouthfeel and afterfeel attributes.

Descriptions given by the QDA panel

Mouthfeel attributeMthick From water to yoghurt thicknessMcreamy Velvety; warm; softMfatty Fatty oily layer in mouth and on lipsMslippery Slippery feeling, smooth, like syrupMsticky Tacky, sticky feelingMdry Dry feeling in the mouth; saliva is absorbed;

swallowing difficultMrough Rough feeling on the teeth and palateMmouthfilling Feeling that whole the mouth is filled upMmelting Thinning of the product in the mouth,

structure disappears

Afterfeel attributeAFastringent Astringent; contracting afterfeelAFdry Saliva absorbing; dry tongueAFrough Rough feeling on the teethAFraw tongue Raw feeling; sandpaper or cat’s tongueAFgrainy Very fine granules; mealy, powdery, floweryAFcreamy Velvety; warm; softAFslimy Feeling of threads at palate; keep swallowingAFcoating Fatty coating on tongue, lips or cheekAFsticky Syrupy; stickilyAFsatiation Hunger alleviation; rich, satisfactoryAFtingling Tingling feeling on lips and/or tongue

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796 791

Nederland BV, The Netherlands) at 10 and 50 times magnification,to check if the droplet size distribution was monomodal and visu-alize droplet aggregation. The particle size distribution (singledroplets and aggregates) was furthermore determined after dilu-tion in water by laser diffractionwith the Mastersizer Hydro 2000S(Malvern Instruments, Southborough, UK). A refractive index of1.469 (sunflower oil) was used to calculate the particle size distri-bution according to Mie theory. Rheological measurements werecarried out in duplicate at 20 �C with a Paar Physica MCR 301 usingthe cone-plate geometry CP75-1, with a cone angle of 1�

(0.0175 rad) and a gap width of 0.05 mm (at the tip). The shear ratewas increased in 40 steps from 1 to 1500 s�1 and subsequentlydecreased to 1 s�1 at the same rate. At each shear rate themeasuring time was 10 s.

2.5. Sensory analysis

2.5.1. MethodThe sensory properties of the emulsions were investigated with

the use of a paid sensory panel trained according to the principlesof quantitative descriptive analysis (QDA) (Stone & Sidel 2004) in atleast four training sessions of 1, 5e2 h (Vingerhoeds, de Wijk, Zoet,Nixdorf, & van Aken, 2008). We chose for QDA because it allows thepanel to use its own terminology to describe all the sensory aspectsof a given set of samples. This reduces the unwanted effect oferroneously collecting sensory differences not available in the list ofsensory attributes pre-defined by the instructor under one of thepre-defined attributes. As a side effect, QDA may generate super-fluous or duplicate attributes, which may be identified andremoved during the training sessions, or show up during statisticalanalysis of the paneling data (van Aken, 2007).

2.5.2. Order of the samplesThe different viscosity levels were evaluated in different sensory

panel sessions. Therefore all samples within one paneling sessionhad similar viscosities and consequently the scorings within onepaneling session are not dominated by large differences in viscositybetween emulsions adjusted to different viscosity levels. Thisallowed a large span of viscosity level variations across the differentpanel sessions, but has the disadvantage that comparison of themagnitude of the viscosity-related attributes (such as Mthick andMcreamy) between viscosity levels is not possible.

2.5.3. SubjectsThe sensory panel consisted of 8 female panellists with an

average age of 42.9 years (standard deviation 9.4 years; range33e59 years). All subjects had previously been screened to excludetaste disorders and to select on good olfactory senses. The panellistsdid not have a scientific background. Testing took place at thesensory facilities of Food and Biobased Research, Wageningen UR,The Netherlands.

2.5.4. ProcedureBefore the paneling, a training session was done to familiarize

the panellists with the set of emulsions to be judged. Panellistswere seated in sensory booths in slightly red light and slight over-pressure to cancel out carrying over effects of odour residuals. Thepanel judged the set of emulsions in a semi-monadic assessmentprocedure in duplicate.

Per session, eight to twelve products were tasted in three blockswith short breaks in between (5e10 min). Each session corre-sponded to one viscosity level. Within each session the order inwhich the emulsions were offered to the panel was randomized.Each session started with a warm-up sample, excluded from thedata analysis. The products were poured in 60 cc cups covered with

a lid, at least 60 min before start of the test. All samples wereoffered at room temperature. The panel members took at leastthree sips (samples without gum arabic) or spoons (samples withgum arabic). For evaluations, the samples were distributedthroughout the mouth for approximately 5 s. The attributes weregenerated and scored in Dutch. For publication, these attributeswere translated to English. During profiling the attributes appearedper category on a monitor placed in front of the panellist with theattributes on the left and a 100-point response scale anchored atthe extremes on the right (Fizz software, Biosystems, v2.21a, 2006).A description of the mouthfeel and afterfeel attributes is given inTable 1. Sensory attributes were assessed in the chronological orderthey are perceived. After the ratings, the emulsions were spat outand afterfeel (AF) attributes were rated. Between each productevaluation, the panellists cleaned their mouth with cream crackers,tap water, and optionally warm tap water.

3. Results

3.1. Emulsion characteristics

Table 2 shows the characteristics of the emulsion systems andthe viscosity curves of viscosity levels B-D are displayed in Fig. 1. Inthe emulsions of viscosity level A (without added gum arabic) thedroplets were unaggregated (the difference between the D32 andD43 values are mainly caused by polydispersity of the droplets).These emulsions seem to be somewhat shear thinning at low shearrates, however this might be due to an artifact in the measurementof the shear stresses, which were very small and became close tothe detection limit for low-viscosity liquids already at low shearrates. At higher shear rates (about 50 s�1 and higher), especially forshear rates that are likely operative in the mouth for these low-viscosity systems (e.g. 500 s�1), these emulsions were nearlyNewtonian. Mixing of the samples with the gum arabic solution ledto a clear aggregation of the emulsion droplets, which was notcompletely reversible on dilution with water and resulted in theincrease inmeasured droplet sizes reported in Table 2. The extent ofaggregationwas especially large for MCTand Castor, and somewhatsmaller for Olive. The precise reason for this aggregation is not

Table 2Sample characteristics.

Viscosity level Oil contentand type of oil

Viscosity (mPa s) Shear thinning index

viscosity at 9:5 s�1

viscosity at 500 s�1

Averaged particlediameter (mm)

at 9.5 s�1 at 50 s�1 at 500 s�1 D32 D43

targeted measured

A 2% MCT 6.05 1a 2.40 1.50 4.03 0.94 1.8220% MCT 7.29 2a 2.62 2.35 3.10 0.96 1.8020% Olive 5.57 2a 3.34 2.28 2.44 1.15 2.4120% Castor 2.69 2a 2.34 2.18 1.22 1.37 3.71

B 2% MCT 180 133 174 163 1.11 0.99 2.5820% MCT 671 133 300 133 5.04 6.74 24.6420% Olive 239 133 152 102 2.25 1.93 41.1720% Castor 250 133 152 98 2.55 6.82 47.75

C 2% MCT 335 335 324 300 1.12 0.93 2.5320% MCT 1212 335 590 280 4.33 6.63 46.8220% Olive 651 335 430 250 2.60 1.72 17.6720% Castor 710 335 377 240 2.96 7.05 44.59

D 2% MCT 4527 2710 2558 2338 1.94 0.93 7.4220% MCT 3311 2710 3018 2190 1.52 4.88 21.4320% Olive 5023 2710 3060 2300 2.18 1.65 17.4920% Castor 3297 2710 2680 1920 1.72 3.84 63.33

a No gum Arabic added.

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796792

known, although depletion flocculation (van Aken, 2006; van Aken,Blijdenstein, & Hotrum, 2003; Asakura & Oosawa, 1954;Blijdenstein, Hendriks, van der Linden, van Vliet, & van Aken,2003; Tuinier & de Kruif, 1999), related to the osmotic pressure ofthe concentrated gum arabic solution, and bridging (gum arabiccontains surface active galactolipids next to polysaccharides) mayplay a role. Probably related to this aggregation, the emulsions weresomewhat more viscous than targeted and shear thinning (Table 2).The magnitude of the viscosity increase and shear thinning effect ishowever not directly correlated to the increases in the apparentparticle size (Table 2), which might be due to the dilution stepinvolved in the droplet size analysis, which moderates depletionflocculation by lowering the osmotic pressure of the gum arabicsolution.

Although the emulsions were not completely Newtonian, thedecrease in viscosity (Fig. 1) between the lowest and highest shearrates in the emulsions thickened with gum arabic remained muchsmaller (about 1 decade or less between 10 and 1500 s�1; shearthinning index typically 2e3) than for similar emulsion systemswith similar viscosities but thickened with xanthan gum, as

0,1

1

10

1 10 100 1000 10000

)s.

aP

(y

tis

oc

siV

Shear rate (s )

Ricine 20 D

MCT 20 D

Olive 20 D

MCT 2 D

Ricine 20 C

MCT 20 C

Olive 20 C

MCT 2 C

MCT 20 B

Ricine 20 B

Olive 20 B

MCT 2 B

Fig. 1. Viscosity versus increasing shear rate for the various samples. The curves arelabeled with the type of oil, oil content (%) and targeted viscosity level at 50 s�1,respectively. Targetted viscosity levels: B: 113 mPa s; C: 335 mPa s; D: 2710 mPa s.

a typical example of a polysaccharide with strongly shear thinningsolutions (more than 2 decades for all viscosity levels with Xan-than; shear thinning index typically 20. Data not shown).

3.2. Sensory results

For each of the targeted viscosity levels, the measured viscosityof the systems varied with the type of oil. As explained above, thiswas probably related to the variation in droplet aggregationwith oiltype. The resulting variation in viscosity is seen to pertain in thesensory scores of especially the textural attributes. For the viscositylevels BeD, the highest correlation between the sensory attributesand viscosity was obtained for the viscosities at low shear rates,measured during an increasing shear rate sweep.

In Figs. 2e4 selected sensory attributes are plotted as a functionof viscosity measured at 9.5 s�1. These attributes were selected onthe basis of significant variation in the sensory paneling results.Fig. 2 shows the results for the mouthfeel attributes Mthick andMcreamy, Fig. 3 for the mouthfeel attributes Mslippery, Mfatty andMsticky, and Fig. 4 for the afterfeel attributes AFcoating andAFcreamy. We remind that because the sensory properties of thesamples were rated in different paneling sessions for the variousviscosity levels, the variation in the sensory ratings cannot bedirectly compared between the viscosity levels (A, B, C, D). Thisexplains why the sensory data from different viscosity levels do notfall on the same curve in each figure.

First concentrating on the 20% oil emulsions with gum arabic(viscosity levels B, C andD), forMthick andMcreamya clear increasein sensory ratings was found as a function of emulsion viscosity(Fig. 2).Msticky, AFcoating andAFcreamy showed a similar increase,but much weaker (Figs. 3 and 4). Instead, Mfatty and Mslippery didnot show any dependence with emulsion viscosity (Fig. 3).

For the 20% fat emulsions without gum arabic (viscosity level A),none of the sensory attributes show a clear relation with theemulsion viscosity (Figs. 2e4), which might be because theviscosity differences between the systems are very small in abso-lute sense. Perhaps the ratings for Mfatty and AFcoating are slightlyhigher for the 20% fat emulsions with Castor compared to the otheroil types for the systems in viscosity level A (Figs. 3 and 4), whichmight be related to the much higher viscosity of Castor. However,this oil-viscosity effect is hardly significant.

0

10

20

30

40

50

60

70

80

90

0,001 0,01 0,1 1 10

k

c

i

h

t

M

A

B C

viscosity (Pa s)

D

0

10

20

30

40

50

60

70

80

0,001 0,01 0,1 1 10

y

m

a

e

r

c

M

A B C D

viscosity (Pa s)

Fig. 2. Mouthfeel attributes Mthick and Mcreamy versus the measured emulsionviscosity at 9.5 s�1. Viscosity of the groups varied as: A: no added thickener; B: targeted133 mPa,s; C: targeted 335 mPa s; D: targeted 2710 mPa s. Symbols: B: MCT; ,:Olive;6: Castor; Open symbol: 2% oil, closed symbols 20% oil. Dotted lines are best fitsthrough the data point at 20% oil content for each of the viscosity groups.

0

10

20

30

40

50

60

70

0,001 0,01 0,1 1 10

y

r

e

p

p

i

l

s

M

A B C

D

viscosity (Pa s)

0

10

20

30

40

50

60

70

80

90

0,001 0,01 0,1 1 10

y

t

t

a

f

M

A B

C D

viscosit y ( Pa s )

0

10

20

30

40

50

60

70

80

90

0,001 0,01 0,1 1 10

y

k

c

i

t

s

M

A

B C

D

viscosity (Pa s)

Fig. 3. Selected textural mouthfeel attributes (Mfatty, Mslippery, Msticky) versus themeasured viscosity at 9.5 s�1. Viscosity of the groups varied as: A: no added thickener;B: targeted 133 mPa s; C: targeted 335 mPa s; D: targeted 2710 mPa s. Symbols: B:MCT; ,: Olive; 6: Castor; Open symbol: 2% oil, closed symbols 20% oil. Dotted linesare best fits through the data point at 20% oil content for each of the viscosity groups.

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796 793

The results show an interesting effect of an increase in oilcontent, as found by increasing theMCTcontent from 2 to 20% (Figs.2e4). For viscosity level A (nogumarabic), the increase in oil contentled to a strong increase in Mthick, Mcreamy, Mslippery, Mfatty,AFcoating and AFcreamy, but not for Msticky. For the systems withgum arabic (viscosity levels BeD), the increase in ratings wasgenerally much smaller. For Mslippery, Mfatty and AFcoating, thedata point for 2% oil is at the same level as thefitted line between thedata points for the emulsionswith 20% oil, in this way correcting forthe effect of the different oils on the emulsion viscosity. For theseattributes the difference in ratings between 2 and 20% MCT istherefore insignificant or absent if we correct for the viscositydifference. Similarly, for Mthick (Fig. 2) the difference might onlyjust be significant for viscosity level B, whereas for Mcreamy thedifference is still significant for viscosity level B (Fig. 2).

Remarkably, forAFcreamy theeffect of increasing theMCTcontentfrom2 to20% remains even for thehigherviscosity levelsBeD(Fig. 4).A completely different behaviour is found for Msticky (Fig. 3). Theincrease in fat content slightly increasesMsticky for systemswithoutgum arabic (viscosity series A), but in contrast decreases for thesystems with gum arabic (viscosity series BeD).

4. Discussion

The results show that the various sensory attributes reactdifferently on the parameters emulsion viscosity, as set by the gum

arabic content, and the fat content. Remarkably, hardly any directeffect was found for the variation in oil viscosity, despite theconsiderable differences in this parameter (by about a factor 30betweenMCTand Castor, which is the largest variation available forconsumable oils). This shows that any oil film that might have beendeposited on the oral surfaces does not significantly contribute tosensory perception by viscous forces generated by shearing of thisfilm. Either such an oil film is not formed (suggesting that theprocess of spreading of the oil content of the emulsion dropletsonto the oral surface through a coalescence-type process does notoccur) or such a film is formed, but its presence is sensed by other

0

10

20

30

40

50

60

70

80

0,001 0,01 0,1 1 10

g

n

i

t

a

o

c

F

A

A B

C D

viscosity (Pa s)

0

10

20

30

40

50

60

70

0,001 0,01 0,1 1 10

y

m

a

e

r

c

F

A

A B

C D

viscosity (Pa s)

Fig. 4. Selected textural afterfeel attributes (AFcoating and AFcreamy) versus themeasured viscosity at 9.5 s�1. Viscosity of the groups varied as: A: no added thickener;B: targeted 133 mPa s; C: targeted 335 mPa s; D: targeted 2710 mPa s. Symbols: B:MCT; ,: Olive; 6: Castor; Open symbol: 2% oil, closed symbols 20% oil. Dotted linesare best fits through the data point at 20% oil content for each of the viscosity groups.

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796794

mechanisms, such as a change in adhesiveness or boundary lubri-cation of the oral surfaces.

For the emulsions with gum arabic, Mthick and Mcreamy areprimarily related to the viscosity of the emulsion system. This wasexpected, since it has been known for long that the Mthick is highlyrelated to the viscosity; for example, a power law dependency isfound for Newtonian liquids (Cutler, Morris, & Taylor, 1983). Simi-larly, it has been demonstrated that also Mcreamy is highlydependent on the viscosity for a broad range of (viscous) foodsystems, although the correlation between Mcreamy and viscositycan be improved by including tribological data (Kilcast & Clegg,2002; Kokini, 1987; de Wijk & Prinz, 2005).

Theweaker dependency of Msticky with the viscosity of the 20%emulsions at viscosity levels B-D shows that the relation withviscosity is less strong. Although one would expect that thestickiness increases with the viscosity (it requires more force andwork to separate two plates if the viscosity of the interspacingliquid is higher), other aspects are also important, such as theadhesion (as for example measured by the contact angle) of theliquid with the surfaces (van Aken, Vingerhoeds, & de Hoog, 2007).Moreover, the effect of the viscosity on Msticky will be complicatedby the interaction with a mucous coating at the oral surfaces.

Mslippery and Mfatty do not show any dependence withemulsion viscosity for the 20% emulsions at viscosity levels BeD.Conceptually, one would expect that Mslippery reflects the lubri-cation of the tongue against the palate in rubbing motion, howeverthe overall rating of the emulsions is rather constant at a neutralvalue of about 50. It might be that this represents the situation ofwell-lubricated oral surfaces (for example by the native mucous

layer or a deposited layer of oil, emulsion droplets or gum arabic).Exceptions are the much lower scoring of Mslippery found for the2% emulsion at viscosity level A and a somewhat decreased valuefor the emulsions at viscosity level D. The first case might representthe situation where the native mucous layer is removed or dilutedwith a solution that is less lubricating because it contains nothickener and only a small amount of oil. The decrease in Mslipperyat highest viscosity level (D) might be caused by the high viscosityin these systems, which will increase the hydrodynamic friction ofthe lubricating layer.

The independence of Mfatty with the emulsion viscositysuggests that the perception of fattiness is not related to viscosity.The observation that Mfatty does not even depend significantly onthe fat content for systems that contain gum arabic (viscosity levelsBeD) suggests that Mfatty is more related to the formation of anadherent coating on the oral surfaces. This is supported by theobservation that the ratings of Mfatty and AFcoating are verysimilar (Figs. 3 and 4). Such a coatingmay consist of a mixture of oil,gum arabic and proteins derived from the mucous layer. Thepossibility of the formation of an oral coating is supported by theobservation by De Jongh and Janssen that food components,especially starch, are retained on the oral surfaces after swallowing(de Jongh & Janssen, 2007).

The behavior of AFcreamy resembles the behaviors of Mfattyand AFcoating, except that for the samples with gum arabic(viscosity levels BeD) AFcreamy is dependent on the fat content. Itmight well be that AFcreamy, like Mfatty and AFcoating, relates tothe sensation of an adherent oral coating of oil, gum arabic andproteins derived from the mucous layer, with the exception thatthis layer should also contain oil.

Altogether, the sensory results of this study show that theemulsified oil has revealed some main sensory effects for liquidemulsions.

1) The emulsion droplets raise the viscosity, acting as a filler. Theincrease in viscosity can be perceived as an increased thicknessand (probably as a consequence) creaminess in systems thatare thickened. If the droplets interact as hard spheres, theviscosity is described up to high volume fractions fe of theemulsion droplets ð0 < fc < 0:5Þ by the semi-empirical Krie-gereDougherty relation (Krieger, 1972),

h ¼ h0

�1� fe

fc

��52fc

(1)

In this equation h0 is the viscosity of the continuous liquid andfc is the volume fraction at random close packing of spheres,which for monodisperse globular particles is a numberbetween 0.58 and 0.64. The viscosity increase becomes evenlarger if the droplets are mutually attractive (Potanin, de Rooij,van den Ende, & Mellema, 1995; Quemada & Berli, 2002; deRooij, Potanin, van den Ende, & Mellema, 1993), by forminga transient structure of droplet aggregates. Because thisstructuring effect is counteracted by shear forces, systems ofmutually attractive droplets are shear thinning.The emulsion droplets also increase the perception of Mthickand Mcreamy in systems without gum arabic (viscosity levelA). This is also expected on the basis of Eq. (1), because alsohere the emulsion droplets increase the viscosity of theemulsion by acting as filler particles. However, unexpectedly,the experimental data do not show a clear relation betweenMthick and Mcreamy and the viscosity of the 20% emulsionswithout gum arabic. Close inspection of the viscosity-sweepdata, revealed that the large difference in viscosity between theCastor (2.658 mPa s) and Olive (5.573 mPa s) measured at

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796 795

9.5 s�1 disappears at higher shear rates, for example at 500 s�1

we found almost equal viscosities of 2.18 mPa s and 2.28 mPa s,respectively. Most likely, at these high shear rates the dropletaggregates are broken down by the shear forces. The experi-mental data therefore suggest that Mthick and Mcreamy areperceived at much higher shear rates for the low-viscositysystems (viscosity level A) than for the systems with gumarabic (viscosity levels BeD). That such a division into twomain regimes for thickness perception indeed exists has beendemonstrated previously by Shama and Sherman (Shama &Sherman 1973a, 1973b). A possible explanation of this divi-sion has been given on the basis of a soft tribology model forthe rubbing motion of a tongue papilla against the palate andthe sensitivity of the mechanoreceptors responsible for tactileperception (van Aken, 2010). This model predicts that thetransition between the two regimes occurs at a viscosity ofabout 100 mPa s and corresponds to a transition between theregimes of boundary friction and hydrodynamic lubrication ofpapilla tips on the tongue. It can be expected that this transi-tion will be gradual in reality, for example stretching overa viscosity range of 10e1000 mPa s. Another aspect that mightbe important for the low-viscosity emulsions is the observationin other studies by our group on very similar emulsion systemsthat the interaction of emulsions with saliva leads to extensivedroplet aggregation (van Aken, Vingerhoeds, & de Hoog, 2005;van Aken et al., 2007; Silletti, Vingerhoeds, Norde, & van Aken,2007a, 2007b; Vingerhoeds, Blijdenstein, Zoet, & van Aken,2005). This aggregation increases the viscosity at low shearrates, and this increase can be considerable. For examplea factor 2e7 was found for an emulsion of 10% oil in waterstabilized by whey protein isolate, which is sufficiently large tobe sensed (van Aken et al., 2007; Silletti, Vingerhoeds, vanAken, & Norde, 2008; Vingerhoeds, Silletti, de Groot,Schipper, & van Aken, 2009). It is conceivable that this saliva-induced droplet aggregation and viscosity increase over-whelms the viscosity differences at low shear rates betweenthe emulsions with different oils and without thickener in thepresent study. It has been shown that saliva addition does notfurther increase the viscosity of an emulsion that has alreadybeen thickened by a polysaccharide (Vingerhoeds et al., 2009),suggesting that saliva-induced aggregation would only beimportant for the emulsions without gum arabic.

2) At low gum arabic content, the emulsion droplets increasea number of textural sensory attributes in a way that is notdirectly related to the viscosity. The descriptors of the attri-butes involved (Mfatty, Mslippery and AFcoating) suggest a roleof the oral surfaces, specifically the formation of a coating.Possibly, the underlying mechanism is that emulsion dropletsare incorporated in the mucous oral coating. Similar ratings ofthese attributes are found for systems with low oil content aslong as they contain gum arabic (levels BeD), suggesting thatthe effect of emulsion droplets on the mucous oral coating canbe taken over by gum arabic (or other thickeners). In thepresence of gum arabic, there is no additional effect of the oilcontent on the attributes involved. The similarity of graphs forAFcoating and Mfatty furthermore suggests that these attri-butes are closely related. Possibly, both are related to thesensation of an oral mucous coating, the difference being thatMfatty is sensed when the emulsion is still in the mouth whileAFcoating is sensed after the product has been swallowed.

3) Finally, it was found that an increase in the fat content leads toan increase in AFcreamy for all systems. This is probably relatedto the formation of an oral mucous coating that contains the oil.The sensed creaminess might be related to an increased

boundary lubrication of the oral mucous coating caused by thepresence of the oil in this layer, in a way that is independent ofthe viscosity of the oil and therefore is not caused by a hydro-dynamically lubricating oil film. Likely, when the tongue isrubbed against the palate, the oil is released from the emulsiondroplets and forms a hydrophobic film that reduces the adhe-sion between themucous layers of opposing oral surfaces. Suchan effect of the oil on the adhesive properties of the oralsurfaces is supported by the observation that an increase in oilcontent led to a reduction in Msticky for systems with gumarabic.

Observations from other studies by our group on very similaremulsion systems confirm the retention of oil in the mouth afterswallowing, strongly supporting the presence of an oral mucouscoating containing emulsion droplets (Dresselhuis, Cohen Stuart,van Aken, Schipper, & de Hoog, 2008), which may spread into anoil film (Dresselhuis, van Aken, de Hoog, & Cohen Stuart, 2008), andlubricate the tongue surface (Dresselhuis et al., 2007; Dresselhuis,van Aken, et al., 2008).

At this point a short discussion of the use of gum arabic as analmost Newtonian representative of the generally muchmore shearthinning group of polysaccharide thickeners is in place. In a parallelstudy, we investigated similar emulsion systems but now thickenedby xanthan gum (unpublished results). As the opposite extreme ofgum arabic, with its highly branched and compact structure insolution, xantan gum solutions consists of extended molecular coilstructures of linear saccharide chains, which very effectivelyincrease the viscosity at low shear rates, but are strongly shearthinning. Because of the much stronger shear thinning behavior ofthe xanthan systems, the viscosity matching of systems withdifferent oil contents to obtain similar viscosities was more critical.For systems that were viscosity-matched at shear rates that arethought to be related to thickness perception in the mouth (around50 s�1 for more viscous systems), the viscosity at high shear rates(for example at 500 s�1) increased with the oil content, because theviscosity-enhancing effect of emulsion droplets as filler particlesremains at the higher shear rates. As a result, viscosity-matchedemulsions with higher fat content were rated to be thicker andcreamier, especially for systems in the low-viscosity range. Apublication on these results is in progress.

5. Conclusions

Based on the presented experimental work and informationfrom previous studies, the present study shows that the effect ofemulsion droplets of liquid oil on the mouthfeel and afterfeel ofliquid systems is determined by three main effects:

1) As the main effect, the emulsion droplets increase the viscosityby acting as filler particles, increasing the volume fraction ofdispersed material. This effect is enhanced by droplet aggre-gation, which may include the effect of saliva-induced dropletaggregation for systems without thickener. The viscosityincrease leads to an increase in thick and creamy mouthfeel. Inthis role, the emulsion droplets can in principle be replaced byother means of increasing the viscosity. For systems witha relatively low-viscosity (below 100 mPa s, for examplea beverage) the use of strongly shear thinning thickeners as fatreplacer will be less successful because the viscosity-relatedattributes are sensed at high shear rates in the mouth.

2) The emulsion droplets can interact with the oral surfaces,which introduce some additional sensory effects. For systemswithout gum arabic, the emulsion droplets are incorporatedinto the mucous oral coating, leading to a retention of the

G.A. van Aken et al. / Food Hydrocolloids 25 (2011) 789e796796

emulsion droplets in the mouth and an increase in semanti-cally-related sensory attributes fatty and slippery mouthfeeland coating afterfeel. A similar sensory effect was obtained bygum arabic, suggesting that this role of emulsion droplets canbe replaced by for example polysaccharide thickeners.

3) An increase in the oil content increases the creamy afterfeelboth in the absence andpresence of gumarabic, and reduces thesticky mouthfeel for systemswith gum arabic. These effects areprobably related to a retention of emulsion droplets within theoral mucous layer. These retained emulsion droplets may forma supply for a deposited oil layer at the frictional contacts. Thesensory effects are independent of the viscosity of the oil, sug-gesting that such an oil film acts like a hydrophobic barrierrather than a hydrodynamic layer. It is not likely that such a filmforming effect can be replaced by a polysaccharide thickener.

References

Akhtar, M., Murray, B. S., & Dickinson, E. (2006). Perception of creaminess of modeloil-in-water dairy emulsions: influence of the shear-thinning nature ofa viscosity-controlling hydrocolloid. Food Hydrocolloids, 20(6), 839e847.

Akhtar, M., Stenzel, J., Murray, B. S., & Dickinson, E. (2005). Factors affecting theperception of creaminess of oil-in-water emulsions. Food Hydrocolloids, 19,521e526.

Asakura, S., & Oosawa, F. (1954). On interaction between two bodies immersed ina solution of macromolecules. Journal of Chemical Physics, 22, 1255e1256.

Blijdenstein, T. B. J., Hendriks,W. P. G., van der Linden, E., vanVliet, T., & vanAken, G. A.(2003). Control of strength and stability of emulsion gels by a combination oflong- and short-range interactions. Langmuir, 19(17), 6657e6663.

Cutler, A. N., Morris, E. R., & Taylor, L. J. (1983). Oral perception of viscosity in fluidfoods and model systems. Journal of Texture Studies, 14, 337e395.

Dresselhuis, D. M., van Aken, G. A., de Hoog, E. H. A., & Cohen Stuart, M. A. (2008).Direct observation of adhesion and spreading of emulsion droplets at solidsurfaces. Soft Matter, 4(5), 1079e1085.

Dresselhuis, D. M., Cohen Stuart, M. A., van Aken, G. A., Schipper, R. G., & deHoog, E. H. A. (2008). Fat retention at the tongue and the role of saliva: adhesionand spreading of ‘protein-poor’ versus ‘protein-rich’ emulsions. Journal ofColloid and Interface Science, 321(1), 21e29.

Dresselhuis, D. M., Klok, H. J., Cohen Stuart, M. A., de Vries, R. J., van Aken, G. A., & deHoog, E. H. A. (2007). Tribology of o/w emulsions under mouth-like conditions:determinants of friction. Food Biophysics, 2, 158e171.

de Jongh,H.H. J., & Janssen,A.M. (2007).Differential clearancekineticsof adhered layerconstituents from the oral cavity asmodulator for afterfeel of dressings: ATR FT-IRmeasurements of localized oral coatings. Journal of Texture Studies, 38, 70e86.

Kilcast, D., & Clegg, S. (2002). Sensory perception of creaminess and its relationshipwith food structure. Food Quality and Preference, 13, 609e623.

Kokini, J. L. (1987). The physical basis of liquid food texture and texture-tasteinteractions. Journal of Food Engineering, 6, 51e81.

Krieger, I. M. (1972). Rheology of monodisperse latices. Advances in Colloid andInterface Science, 3, 111e136.

Nishinari, K. (2006). Polysaccharide rheology and in-mouth perception. InA. M. Stephen, G. O. Phillips, & P. A. Williams (Eds.), Food polysaccharides andtheir applications (pp. 541e588). Boca Raton FL: Taylor & Francis.

Potanin, A. A., de Rooij, R., van den Ende, D., & Mellema, J. (1995). Microrheologicalmodeling of weakly aggregated dispersions. Journal of Chemical Physics, 102,5845e5853.

Quemada, D., & Berli, C. (2002). Energy of interaction in colloids and its implicationsin rheological modeling. Advances in Colloid and Interface Science, 98, 51e85.

de Rooij, R., Potanin, A. A., van den Ende, D., & Mellema, J. (1993). Steady shearviscosity of weakly aggregating polystyrene latex dispersions. Journal ofChemical Physics, 99(11), 9213e9223.

Shama, F., & Sherman, P. (1973a). Identification of stimuli controlling the sensoryevaluation of viscosity, II oral methods. Journal of Texture Studies, 4, 111e118.

Shama, F., & Sherman, P. (1973b). Variation in stimuli associated with oral evalua-tion of the viscosities of glucose solutions. Journal of Texture Studies, 4(2),254e262.

Silletti, E., Vingerhoeds, M. H., van Aken, G. A., & Norde, W. (2008). Rheologicalbehaviour of food emulsions mixed with saliva: effect of droplet size, oilcontent, salivary protein content and saliva type. Food Biophysics, 3, 318e328.

Silletti, E., Vingerhoeds, M. H., Norde, W., & van Aken, G. A. (2007a). Complexformation in mixtures of lysozyme-stabilized emulsions and human saliva.Journal of Colloid and Interface Science, 313(2), 485e493.

Silletti, E., Vingerhoeds, M. H., Norde, W., & van Aken, G. A. (2007b). The role ofelectrostatics in saliva-induced emulsion flocculation. Food Hydrocolloids, 21(4),596e606.

Stone, H., & Sidel, J. L. (2004). Sensory evaluation practices (3rd ed.). New York:Elsevier Academic Press.

Tuinier, R., & de Kruif, C. G. (1999). Phase separation, creaming, and networkformation of oil-in-water emulsions induced by an exocellular polysaccharide.Journal of Colloid and Interface Science, 218(1), 201e210.

van Aken, G. A. (2006). Polysaccharides in food emulsions. In P. A. Williams (Ed.),Food polysaccharides and their application (2nd ed.). (pp. 519e537) New York:Marcel Dekker.

van Aken, G. A. (2007). Relating food microstructure to sensory quality. InD. J. McClements (Ed.), Understanding and controlling the microstructure ofcomplex foods (pp. 449e482). Abington, Cambridge, UK: Woodhead Publishing.

van Aken, G. A. (2010). Modelling texture perception by soft epithelial surfaces. SoftMatter, 6, 826e834.

van Aken, G. A., Blijdenstein, T. B. J., & Hotrum, N. E. (2003). Colloidal destabilisationmechanisms in emulsions stabilised by proteins. Current Opinion in Colloid andInterface Science, 8, 371e379.

van Aken, G. A., de Hoog, E. H. A., Nixdorf, R. R., Zoet, F. D., & Vingerhoeds, M. H.(2006). Aspects of sensory perception of food emulsions thickened by poly-saccharides. In P. A. Williams, & G. O. Phillips (Eds.), Gums and stabilisers for thefood industry, Vol. 13 (pp. 449e456). Cambridge: The Royal Society of Chemistry,RSC special publication 303.

van Aken, G. A., Vingerhoeds, M. H., & de Hoog, E. H. A. (2005). Colloidal behaviourof food emulsions under oral conditions. In E. Dickinson (Ed.), Food colloids2004: interactions, microstructure and processing (pp. 356e366). Cambridge: TheRoyal Society of Chemistry.

van Aken, G. A., Vingerhoeds, M. H., & de Hoog, E. H. A. (2007). Food colloids underoral conditions. Current Opinions in Colloid and Interface Science, 12(4e5),251e262.

van Aken, G. A., Vingerhoeds, M. H., & de Hoog, E. H. A. (2009). Oral processing andperception of food emulsions: the relevance for fat reduction in food. InD. J. McClements, & E. A. Decker (Eds.), Designing functional foods (pp. 481e501).Cambridge, UK: Woodhead Publishing Ltd.

Vingerhoeds, M. H., Blijdenstein, T. B. J., Zoet, F. D., & van Aken, G. A. (2005).Emulsion flocculation induced by saliva and mucin. Food Hydrocolloids, 19,915e922.

Vingerhoeds, M. H., Silletti, E., de Groot, J., Schipper, R. G., & van Aken, G. A. (2009).Relating the effect of saliva-induced emulsion flocculation on rheologicalproperties and adherence to the tongue surface with sensory perception. FoodHydrocolloids, 23(3), 773e785.

Vingerhoeds, M. H., de Wijk, R. A., Zoet, F. D., Nixdorf, R. R., & van Aken, G. A. (2008).How emulsion composition and structure affect sensory perception of lowviscosity model emulsions. Food Hydrocolloids, 22(4), 631e646.

de Wijk, R. A., & Prinz, J. F. (2005). The role of friction in perceived oral texture. FoodQuality and Preference, 16, 121e129.

Wood, F. W. (1968). Psychophysical studies on the consistency of liquid foods. Rheologyand texture of foodstuffs. In SCI Monograph No. 27 (pp. 40e49). London: Societyof Chemical Industry.