Chemosensory Perception

, Volume 4, Issue 1, pp 9–15

Enhancement of Saltiness Perception in Hyperosmotic Solutions


    • Division of Food SciencesUniversity of Nottingham (UNott)
  • Christophe Michon
    • Division of Food SciencesUniversity of Nottingham (UNott)
  • Cécile Morris
    • Division of Food SciencesUniversity of Nottingham (UNott)
  • Louise Hewson
    • Division of Food SciencesUniversity of Nottingham (UNott)
  • Joanne Hort
    • Division of Food SciencesUniversity of Nottingham (UNott)
  • Andrew J. Taylor
    • Division of Food SciencesUniversity of Nottingham (UNott)
    • Division of Food SciencesUniversity of Nottingham (UNott)

DOI: 10.1007/s12078-011-9083-7

Cite this article as:
Koliandris, A., Michon, C., Morris, C. et al. Chem. Percept. (2011) 4: 9. doi:10.1007/s12078-011-9083-7


Salt (sodium chloride) plays a major role in perception of flavor in food products. Though reducing sodium content in processed food could significantly improve the health level of the population, the detrimental change in flavor presents a major challenge as consumers generally find low salt products unacceptable. Therefore, technological solutions are being sought to lower the salt content of processed foods without altering their taste. In order to better understand saltiness perception in thickened products such as soups and sauces, this study was designed to evaluate the possibility of enhancing saltiness perception through use of hyperosmotic solutions containing high polymer concentration (up to 30%). Saltiness and sweetness perception were investigated in Newtonian solutions of identical viscosity thickened with different concentrations of dextran, which was achieved by using dextrans of different molecular weights. Attribute difference tests (paired comparisons and multiple paired comparisons) were performed by untrained subjects. A significant enhancement of saltiness, but not of sweetness, was found in hyperosmotic solutions (higher polymer concentration) compared to solutions of lower osmolality (lower polymer concentration). The present results may be considered as a human study validation of an in vitro demonstrated effect of osmolality on the response of taste receptor cells to NaCl and suggests that high concentrations of low molecular weight thickeners could be used to enhance saltiness perception in low salt products.


Food ThickenerOsmolalitySalt ReductionSensory AnalysisSodium ChlorideTasteViscosity


In most developed countries, the Food Regulatory Bodies have set targets to reduce sodium chloride consumption and therefore decrease the incidence of hypertension, the leading cause of preventable illness and death after smoking (He and MacGregor 2010). As about 80% of salt (sodium chloride) intake originates from processed foods (Angus 2007), the food industry has been urged to reduce salt in food formulations. This has an impact on flavor perception and affects consumers’ preference, which is why technological solutions may be sought to enhance saltiness perception. For semi-liquid products, the fact that saltiness perception decreases with increasing viscosity has been known since the 1950s (Mackey 1958; Pangborn and Trabue 1973; Christensen 1980; Baines and Morris 1987). It may be exploited to optimize saltiness perception by careful choice of a food thickener and the flow behavior of the product. Recently, it has been reported that saltiness perception in semi-liquid products is not only affected by the low shear viscosity of the solution, but also by the concentration of the polymer added to obtain desired product viscosity (Koliandris et al. 2010). It was hypothesized that the taste enhancement at high polymer concentration was due to the high osmolality of the samples. This hypothesis was tested and the results are reported here.

Osmolality is defined as the number of moles of solute particles in a given mass of solvent and its unit is Osm kg−1. Osmolality is different from molality as it measures moles of solute particles rather than moles of solute, and some solutes dissociate into several particles (two particles, Na+ and Cl, in case of NaCl). It is well known that the osmolality of the medium surrounding a living cell affects the cell volume, as water passes through the cell membrane from the compartment of low osmolality to the compartment of high osmolality (Evans 2008). The imbalance in osmolality between inner and outer cell compartment is either temporary or permanent depending on whether the solutes can or cannot permeate through the cell membrane.

As with any other cell, taste receptors cells (TRC) experience changes in osmotic pressure as the osmolality of saliva is about 150 mOsm kg−1 and the osmolality of many foods and beverages ranges from nearly 0 (water) to more than 2,500 mOsm kg−1 (wine) (Feldman and Barnett 1995). Therefore, it has been suggested that solution osmolality may be an important variable to consider when determining the overall gustatory response of a food or beverage. Detection of the low osmolality of water by the TRCs may be at the origin of the recognition of water taste in mammals (Gilbertson 2002) or insects (Cameron et al. 2010; Chen et al. 2010). Lyall et al. (1999) observed an increase in neural response of rats to NaCl solution when mannitol or cellobiose were also present in the solution, rendering it hyperosmotic. This was associated with a sustained decrease in TRC volume. Such an effect was inactivated by the presence of amiloride, an agent blocking the epithelial sodium channel involved in saltiness perception (DeSimone and Lyall 2008). However, in contrast to rodents, amiloride-sensitive sodium channels are not the predominant transducer of salt taste in humans (Ossebaard and Smith 1995). Amiloride-insensitive Na+ channels present in the human sub-mucosa are also involved in human salt taste perception, and solutes have to diffuse through the “tight junctions” linking the TRCs to reach these channels (Ye et al. 1991; Stewart et al. 1997; Michlig et al. 2007).

In the study reported here, the effect of polymer concentration on taste perception was investigated to answer the question whether the level of osmolality may be responsible for taste enhancement found previously for highly concentrated biopolymer solutions (Koliandris et al. 2010). For this purpose, solutions were thickened with dextran, a linear polymer composed of α−1→6 glucose exhibiting Newtonian flow behavior in solution (Sabatie et al. 1988; Tirtaatmadja et al. 2001). A range of molecular weights was used allowing preparation of constant viscosity solutions at various levels of osmolality to study the effect of polymer concentration on sweetness and saltiness perception. These two taste modalities were chosen as they have been reported to be the most affected by the addition of thickeners (Lynch et al. 1993; Cook et al. 2002). In addition, the mechanisms of sweetness and saltiness perception are very different. Indeed, the sweet taste modality is mediated by G-protein-coupled receptors, whereby the tastant binds to the extracellular domain of the receptor. In contrast, salty tastants modulate the taste cell function by direct entry of sodium in the cell, through a specialized ionic channel. Hence, osmolality is expected to affect sweet and salty taste modalities in a different way.

Materials and Methods


Three dextran samples (clinical-grade dextran, Pharmacosmos, Holbaek, Denmark) with a nominal molecular weight of 10, 40, and 500 kDa were used. The sodium, potassium, and protein contents were checked against specifications prior to the main trials. The residues were found to be low enough (proteins < 0.13%, Na < 0.011%, K < 0.01% of the dextran powders) to be deemed negligible. Sodium chloride and sucrose were added as “Table salt” and “sugar” (bought in a local supermarket). Aqueous solutions were prepared with bottled water (Evian, Danone, Evian-les-Bains, France) since the samples were also designated for human consumption. “Evian” was chosen for its low ion content; in the following, it is simply referred to as “water”.

Solution Preparation and Rheological Properties

Samples were prepared by initially dispersing the dry powder in water using a magnetic stirrer followed by gentle mechanical agitation in an end-over-end mixer (Reax2, Heidolph, Kelheim, Germany). Dissolution occurred at room temperature and was complete after 3 h. Next, tastant molecules were added at the desired concentration (weight/volume basis) followed by further mixing at room temperature in the end-over-end mixer for 2 h. Samples were stored at 4°C and used within 2 days.

All rheological measurements were conducted at 20°C using a rotational rheometer (MCR301, Anton Paar, Graz, Austria). Steady state viscosity data were acquired using a smooth cone-plate geometry (50 mm, 2°) applying shear rates up to 103 s−1. The extensional viscosity of the samples was evaluated with a capillary break-up extensional rheometer (CaBER 1, Thermo Haake, Karlsruhe, Germany). The samples exhibited largely inelastic behavior and the molecular weight of the dextrans used did not induce significant differences in extensional viscosity, thus data are not shown.

Osmolality was measured in triplicate with an osmometer using the depression of freezing point method (Advanced® Model 3300 Micro-Osmometer, Advanced Instruments, Norwood, USA).


Solutions of different molecular weight dextran matched in viscosity at 30 ± 2 mPa s at 20°C were prepared by adjusting polymer concentration accordingly. The viscosity chosen and the range of sodium chloride concentration (0.45–0.6%, see below) are typical of a low-viscosity soup. Solution concentrations were 10.25%, 20.5%, and 30% w/w for 500 kDa, 40 and 10 kDa dextran, respectively, and in the following the samples are referred to as “L”, “M” and “H”, respectively for “Low”, “Medium” and “High” polymer concentration, respectively. Additionally, in order to obtain samples matched in osmolality, two samples exhibiting a different viscosity level (9 and 50 mPa s) were prepared for one of the sensory analyses (test D2) using dextran 500 and dextran 10 kDa.

Four sensory analysis tests were carried out as described in the following section. The objective of these tests was to identify whether saltiness perception (test B) and sweetness perception (test C) differed across samples. A preliminary test (test A) was carried to ensure that samples could not be distinguished in absence of tastant. Finally, the saltiness perception in samples of matched osmolarity was investigated in test D.

Sensory Assessment

Volunteers were recruited from staff and students (aged 21–58, ±50% women depending on the test) to take part in the studies. Although no specific training was given, all volunteers had prior experience of the sensory methodology used (either triangle tests or paired comparison tests). The study was approved by the local Ethics Committee. Tests were conducted in individual booths, lit with northern hemisphere lighting, in a quiet, air-conditioned room (20°C). Still mineral water (Evian) and unsalted crackers (99% fat free, Rakusen’s, Leeds, UK) were provided as palate cleansers. All samples were coded with random three digit numbers and were presented at room temperature on a disposable spoon (2 ml). Panelists were instructed to take the whole sample in their mouth, to evaluate the sensory attributes, and then to swallow it. No further instructions regarding manipulation in the mouth were given. Data were collected using the computerized data acquisition system Fizz (Biosystèmes, Couternon, France).

Test A: No Tastant Added

As a preliminary test, it was important to ensure that subjects could not distinguish the samples based on the intrinsic taste of the thickeners or on differences in mouthfeel (in spite of the identical shear viscosity). Therefore, an overall difference test (triangle test, ISO 4120:2004), was deemed the most appropriate sensory assessment for this part of the study. Three triangle tests were performed on each possible pairing of samples L, M, and H with no added tastant (L versus M, M versus H, and L versus H). Subjects (N = 27) were presented with three coded samples in a specific order whereby two were identical and one was different. They were instructed to taste (also smell and examine) the samples and to identify the odd sample. Data was analyzed using Fizz software (α = 0.05) according to ISO 4120:2004.

Test B: Effect of Polymer Concentration on Saltiness Perception

In order to investigate the effect of polymer concentration on saltiness perception, a fixed amount of sodium chloride (0.6% w/v) was added to samples L, M, H. The addition of sodium chloride did not modify the shear viscosity (results not shown). The salt containing samples were compared for saltiness using multiple paired comparisons (test B1): all three possible pairings of the three solutions L, M, and H (L versus M, M versus H, and L versus H) were evaluated in one panel session. Subjects (N = 33) assessed each pair of solutions only once; presentation order of samples was randomized and balanced across the panel, based on a Latin square design. Subjects were instructed to indicate (forced choice mode) the saltier sample of each pair. Panel judgments on each pair of samples were pooled and tabulated. The results were subjected to Friedman analysis (Meilgaard et al. 2006). The rank sums were calculated by adding the sum of the row frequencies to twice the sum of the column frequencies. The calculation of rank sums by the Friedmann analysis allows to visualize the results on a single dimension (rank sums) and to check for inconsistencies. Rank sums were presented on a line scale for relative saltiness. Significant differences between samples were identified by calculating the honestly significant different value for comparing two rank sums (Meilgaard et al. 2006).

To extend the results of test B1, the two extreme samples L and H were prepared at a lower sodium chloride level (0.45% w/v) and their perceived saltiness was again compared using paired comparison (ISO 5495:2005; test B2): subjects (N = 40) were presented with a pair of samples and were asked to judge (forced choice mode) which solution tasted the saltier.

Test C: Effect of Polymer Concentration on Sweetness Perception

The objective of test C was to determine whether sweetness perception was affected by the polymer concentration similarly to saltiness perception. A fixed amount of sucrose (5% w/v) was added to samples L, M, and H and they were compared for sweetness perception using multiple paired comparison (three tests: L versus M, M versus H, and L versus H). Data acquisition and analysis was performed as indicated for test B1 and 27 subjects participated in the test. The addition of sucrose at the level used in this study did not significantly modify shear viscosity (results not shown).

Saltiness Perception in Samples of Identical Osmolality (Test D)

A further set of samples was prepared to compare the sensory properties when osmolality was matched using different dextrans. In test D, osmolality was matched between two samples thickened with medium and high molecular weight dextran (40 and 500 kDa, respectively). Since sample osmolality, in case of the samples used in study, is a function of salt concentration and polymer concentration (which in turn determines the viscosity), matching osmolality implies that samples either differ in sodium chloride concentration (test D1) or in viscosity (test D2). As shown in Table 1, two samples of identical osmolality and viscosity, but different sodium chloride concentration were compared in Test D1. In test D2, sodium chloride concentration was constant but viscosity of the samples differed since polymer concentration was adjusted to match osmolality. The two samples within each test were compared using paired comparison tests (ISO 5495:2005): subjects (N = 42) were presented with a pair of samples and were asked to judge (forced choice mode) which solution tasted the saltier.
Table 1

Composition of the samples for test D1 and D2 and results of the paired comparison test


40 kDa dextran

500 kDa dextran

Test D1

Osmolality (mOsm kg−1)

290 ± 5

290 ± 5

Viscosity (mPa s)

30 ± 2

30 ± 2

Polymer (% w/w)

20.5 (as sample M in tests A, B, C)

10.25 (as sample L in tests A, B, C)

Sodium chloride (% w/v)



Number of subjects judging it saltier (total 42 panelists)


p < 0.001


Test D2

Osmolality (mOsm kg−1)

281 ± 3

281 ± 3

Viscosity (mPa s)



Polymer (% w/w)



Sodium chloride (% w/v)



Number of subjects judging it saltier (total 42 panelists)


p < 0.01



Sample Perception in Absence of Tastant (Test A)

Equi-viscous samples thickened with a high concentration of low molecular weight dextran (sample H), a medium concentration of medium molecular weight dextran (sample M) or a low concentration of high molecular weight dextran (sample L) were compared with no added tastant using triangle tests. As reported in Table 2, large differences in osmolality exist between samples with L exhibiting the lowest osmolality (33 mOsm kg−1) and H the highest (431 mOsm kg−1). Comparing L and M, 10 out of 27 subjects correctly identified the odd sample (p = 0.266). With regard to L and H, 11 out of 27 subjects found the odd sample (p = 0.411), and 9 out of 27 subjects correctly identified the odd sample (p = 0.57) when comparing M to H. A minimum of 14 subjects finding the odd sample would have been required to infer that sensory perception of the samples was significantly different (p = 0.05). Thus, it can be concluded that in absence of tastant, sensory differences due to dextran molecular weight and concentration, if they exist, are not sufficient to enable discrimination of the samples.
Table 2

Osmolality of dextran samples of different molecular weight, viscosity matched at 30 mPa s in absence or presence of tastant

Sample code and description

Osmolality (mOsm kg−1)

No tastant

0.6% w/v salt

5% sucrose


10.25% w/v 500 kDa dextran





20.5% w/v 40 kDa dextran





30.0% w/v 10 kDa dextran




All results are significantly different (p < 0.001). “L” stands for low polymer concentration (associated with a low osmolality) “M” for medium polymer concentration (medium osmolality) and “H” for high polymer concentration (high osmolality)

Effect of Polymer Concentration on Saltiness Perception (Test B)

Sodium chloride at a level of 0.6% w/v was added to the equi-viscous dextran solutions of different polymer concentration (samples L, M, and H) and subjects were instructed to compare the samples for saltiness using multiple paired comparisons (test B1). The rank sums are depicted in Fig. 1. Subjects were able to distinguish the samples based on saltiness (Friedman’s statistic T = 13.6, p < 0.01). L appeared significantly less salty than M (as found by 21 of the 33 subjects) and M was significantly less salty than H (as found by 21 of the 33 subjects). H was found to be significantly saltier than L (by 25 of the 33 subjects, p < 0.01). As expected, an increase in the sample osmolality was observed with increasing polymer concentration (see Fig. 1 or Table 2) demonstrating the correlation between saltiness perception and sample osmolality.
Fig. 1

Line diagram representation the rank sum scores for saltiness in the multiple paired comparisons of dextran samples L (10.25% of dextran 500 kDa, low polymer concentration, low osmolality), M (20.5% dextran 40 kDa, medium polymer concentration, medium osmolality) and H (30% of dextran 10 kDa, high polymer concentration, high osmolality) containing 0.6% w/v sodium chloride (viscosity 30 mPa s, 33 subjects) and osmolality of the samples. For rank sums, samples with the same letter code are not significantly different (p < 0.1). L and H are significantly different at p = 0.001. Values for osmolality indexed with *** are significantly different at p = 0.001

As shown in Table 3, the effect of polymer concentration on saltiness perception was also investigated at a lower concentration of sodium chloride (0.45% w/v, test B2). Sample H exhibited high osmolality (650 mOsm kg−1) due to the higher polymer concentration and was found significantly saltier than the low osmolality sample L (196 mOsm kg−1) (p = 0.006).
Table 3

Results of paired comparison test B2 between sample L (10.25% of dextran 500 kDa, low polymer concentration, low osmolality) and H (30% of dextran 10 kDa, high polymer concentration, high osmolality) containing 0.45% w/v sodium chloride (40 panelists)


Sodium chloride (% w/v)

Osmolality (mOsm kg−1)

Number of times found saltier




196 ± 5




650 ± 20


Effect of Polymer Concentration on Sweetness Perception (Test C)

To investigate whether polymer concentration impacts on sweetness perception in the same way as it impacts on saltiness perception, the same samples as used in Test B1 (viscosity matched (30 mPa s) solutions of dextran of three different molecular weights) but containing 5% w/v sugar instead of sodium chloride were compared for sweetness using multiple paired comparisons (test C). The rank sums are depicted in Fig. 2. Results indicated no statistical evidence to conclude significant differences in perceived sweetness between the samples (Friedman’s statistic T = 3.8, p > 0.1), which contrasts with the saltiness results of tests B1 and B2.
Fig. 2

Line diagram representation of the rank sum scores for sweetness in the multiple paired comparisons of dextran samples L (10.25% of dextran 500 kDa, low polymer concentration, low osmolality), M (20.5% dextran 40 kDa, medium polymer concentration, medium osmolality), and H (30% of dextran 10 kDa, high polymer concentration, high osmolality) containing 5% w/v sugar (viscosity 30 mPa s, 27 subjects) and osmolality of the samples. For rank sums, samples with the same letter code are not significantly different (p < 0.1). Values for osmolality indexed with *** are significantly different at p = 0.001

Saltiness Perception in Samples of Identical Osmolality (Test D)

In test D, samples matched in osmolality, thus varying either in sodium chloride concentration (test D1) or viscosity (test D2), were evaluated and the results are shown in Table 1. In test D1, the sample containing the lower molecular weight dextran (40 kDa) appeared significantly less salty (p < 0.001). This is not surprising as the difference in sodium chloride concentration between the two samples (0.3% w/v) was much higher than the just noticeable difference (about 0.058% for a level of sodium chloride close to 0.8%) (Johansson et al. 1973). The results of test D2 are clearly a reflection of the effect of viscosity on saltiness perception as the sample containing the low molecular weight dextran, exhibiting a much lower viscosity (9 vs. 50 mPa s), was found to be perceived as significantly more salty (p < 0.01).


The enhancement of saltiness perception with increasing osmolality (increased as a result of increasing polymer concentration) was verified for dextran solutions at two concentrations of sodium chloride (0.6% and 0.45% w/v). The results corroborate previous findings on the same polymer solution system (Koliandris et al. 2010). Conclusive research on the effect of polymer concentration on saltiness perception has to the best of the authors’ knowledge not previously been published as typically highly efficient thickeners are used (e.g., carboxymethylcellulose, xanthan gum, guar gum, carrageenan, HPMC, methylcellulose) and, therefore, polymer concentrations applied tend to be low (max. 2%) not covering a sufficiently broad range to reveal effects of polymer concentration on saltiness perception. In addition, these polymers are shear-thinning, so that effects of polymer concentration would have been difficult to separate from effects of variable shear viscosity.

Dextrans of different molecular weight may generate different “off-tastes” and, though the shear and extensional viscosity of the samples used here were identical, the differences in polymer concentration may have generated differences in mouthfeel. In the absence of a tastant (test A), subjects were not able to discriminate between the samples. Thus, differences found for saltiness perception did not arise from a stimulus error, where panelists used an irrelevant criterion (e.g., mouthfeel) to distinguish between samples.

Solution salt concentrations were controlled on a w/v basis, and one could argue that simply the increase in sodium chloride concentration based on the aqueous phase (grams per 100 g of water) upon increase in polymer concentration (adjusted on a w/w basis) was the origin of enhanced saltiness perception. Indeed, for a set concentration of 0.6% w/v of NaCl, a 10% w/w dextran solution contains 0.64 g of NaCl for 100 g water whereas a 30% w/w dextran solution contains 0.76 g NaCl for a 100 g of water. However, if the stimulus for taste perception is the tastant concentration per grams of water, rather than per grams (w/w basis) or per liters (w/v basis) based on the total system, one would have expected to also find an enhancement of all taste modalities with increasing polymer concentration. This was not the case for sweetness (test C), thus, the most likely origin of the enhanced saltiness perception lies in the increased osmolality of these samples and the mechanism of salt taste perception.

Based on in vitro studies on rats, Lyall et al. (1999) have previously reported an osmolyte-induced increase in rats’ chorda tympani response to NaCl involving the activation of apical Na+ channels. NaCl activates these ion channels and, therefore, the high osmolality of the samples may have increased the nervous response, resulting in higher saltiness perception observed in test B. In contrast, sugar does not activate Na+ channels, which is congruent with the absence of effect of solution osmolality on sweetness perception reported here (test C). As described in the introduction, it was demonstrated in vitro that the effect of osmolality involved amiloride-sensitive ion channels (Lyall et al. 1999), yet in humans this pathway is not predominant (Ossebaard and Smith 1995). The present data indicate that a modulation of the activity of the amiloride-sensitive ion channel would enable control over saltiness perception. Alternatively, it may be hypothesized that in human, osmolality also affects the submucosal pathway for saltiness transduction. For example, it is possible that the decrease in cell volume observed in hyperosmotic solutions facilitates the passage of salt ions through the tight junctions, thereby increasing the activity of the submucosal ion channels.

It is worth stressing that the osmolality effect on taste perception depends on the taste modality studied, which may explain some of the taste-specific effects of thickeners on taste perception reported in the literature (Moskowitz and Arabie 1970; Pangborn and Trabue 1973).The present results are based on a model system for soups with relevant salt levels, further studies may include broadening the concentration range from near recognition threshold to strong intensity and different osmotic substances may be investigated.


In the research reported, taste perception was investigated in dextran solutions of identical shear viscosity and varying polymer concentration which induced large differences in osmolality. Polymer concentration was found to significantly affect saltiness perception, but not sweetness perception. The results show that osmolality affects saltiness perception. Food products of high osmolality could be designed through careful choice of molecular weight and concentration of thickener to enhance saltiness perception in low salt products. However, the use of higher concentrations of thickeners is likely to be associated with increased cost of recipe.


This research was part of a project funded by the Technology Strategy Board’s Collaborative Research and Development program (TP/6/DAM/6/S/K3004C).

Copyright information

© Springer Science + Business Media, LLC 2011