Possible Mechanism behind the Hard-to-Swallow Property of Oil Seed Pastes

Roasted and crushed oil-rich seeds, such as sesame paste and peanut butter, both share a common structure and elicit an apparent sensation of thickening in the mouth. Working with sesame paste, as an example, the force needed to compress sesame paste:water mixtures peaked at 25% added water. The adhesive force required to pull a plunger from the surface was bimodal with peaks at around 15 and 25% hydration. It is postulated that when introduced to the mouth, water from the saliva is absorbed by the paste leading to a hard, adhesive material that sticks to the palate and the tongue, making these materials hard to swallow. It is hypothesized that the shared hard-to-swallow behaviour exhibited by other oil seed pastes/butters is due to a similar hydration process in the mouth.


INTRODUCTION
Dried, roasted/toasted, and ground seeds from peanuts, cashew, sunflower, hazelnuts, and sesame are often used as spreads or ingredients in the cuisine of various cultures around the world. [1−4] Products, such as peanut butter, tahini, hazelnut butter, cashew nut butter, and sunflower spread, share a common structure, consisting of a fragments of cell debris suspended in oil. Anecdotal observations from people who eat these products, suggest that they appear to thicken in the mouth. Rosenthal and Share [5] coined the phrase "hard to swallow oil seed pastes" to describe the experience. Such a behaviour is not unlike dilatancy (shear thickening), which is exhibited by other concentrated suspensions, such as starch granules in water, and which may be responsible for this unusual mouthfeel.
The authors refer to this phenomenon as "anecdotal" because there have been relatively few scientific publications on the matter; however, Chen and Lolivert [6] investigated the swallowing time of 28 paste-like and liquid foods (their premise was that liquid foods would not require mastication for particle size reduction), and they found that of the foods that they considered smooth peanut butter was the most difficult to swallow, with an average oral residence time of 7 (±1.7) s. More recently, the temporal dominance of sensation technique has been applied to compare the sensations during oral processing of whole peanuts, peanut meal, and peanut paste. It shows that, while the oral trajectory for whole peanuts and peanut meal starts hard and crunchy, they gradually become compacted on the teeth. In contrast, peanut paste, which possesses relatively little structure, begins to thicken and stick to the palate soon after entering the mouth. [5] This article seeks to investigate the mechanism behind this hard-to-swallow phenomenon using sesame paste as a model, with the hope that the findings may be transferable to other similarly structured materials, which seem to exhibit this hard-to-swallow phenomenon.
Sesame paste (tahin, tahina, tehina, tahini) is widely used in middle eastern cuisine. The paste may be eaten on its own or as an ingredient in savoury dishes, such as Humous (with chick peas, lemon juice and garlic), Babaganoush (with aubergine, garlic), as well as sweet dishes like Halva (with syrups, honey, or sugar). It is derived by grinding the dried, roasted, oil rich seed of Sesamum indicum, and it yields an oily, buff-brown coloured, opaque liquid consisting of a suspension of cell debris in expressed oil.
From its paste-like appearance one might expect that tahini would be easy to swallow, yet as mentioned above it exhibits the hard-to-swallow phenomenon when eaten on its own. The suggestion that sesame paste might be shear thickening may be discounted as several studies have looked at its rheology, showing that it exhibits a strong shear thinning (pseudoplastic) behaviour, [7−9] which is sometimes thixotropic. [10] Another possibility is that the sesame paste is reacting with saliva in the mouth, resulting in changes in the rheological behaviour. Lindner and Kinsella [11] investigated the hydration of sesame paste and showed a rise in viscosity as the water content rose to about 12%. Above 12% the consistency mixture starts to solidify and further viscosity measurements were not possible until the water content reached levels of 40% and above, when an oil-in-water emulsion is formed. This rise in apparent viscosity caused by hydration may be partially responsible for the hard-to-swallow sensation, but no data exists for what happens between 12 and 40% hydration. This article seeks to investigate the changes in rheological behaviour within this hydration range.
To monitor this change in viscoelastic behaviour, we chose to undertake uniaxial compression of the mixture by forcing a flat ended probe into the mixture while measuring the force needed to penetrate the surface. The adhesive force of the material is also measured by measuring the tack while withdrawing the probe after the compression. [12] Obviously, adding water to a dry material suspended in oil will not thin the dispersion. As would be expected, the initial effect is for the dry cell debris to become hydrated while still suspended in oil. In order that we could gauge the interaction of the water and the sesame paste constituents, we also examined the water activity of the mixtures-thus giving a sense of how added water associates with the diverse mixture materials likely to be present in the cell debris. The objective of this study was to fill gaps in the physical data previously collected on the hydration properties of tahini and to relate this to reports of the oral behaviour of sesame paste and other similar products formed by crushing toasted oil-rich seeds.

METHODOLOGY Sample Preparation
Sesame paste (Al Taj, Riyadh, Kingdom of Saudi Arabia) was stirred to ensure a homogeneous composition. For the water activity and uniaxial compression, mixtures of sesame paste and distilled water were prepared on a weight-weight basis to achieve the following percentages of added water: 0, 2, 4, 5, 6, 8, 10, 12, 14, 15, 16, 18, 20, 22, 24, 25, 26, 28, 30, 32, 34, 35, 36, 38, 40, and 50% water. The mixtures were mixed well with a spatula and about 10 g of each mixture was placed into three separate 40-mm-diameter, water activity dishes (Novasania, Switzerland), the surface of the mixture was smoothed out to fill the dish with a plane surface. The dishes were lidded and left in the dark at room temperature for two days in order for the water distribution to equilibrate (the 40 and 50% added water samples were stored at 4 • C to prevent mould growth and then warmed to room temperature prior to further testing).

Measurement of Water Activity
A water activity meter (Novasania TH200) was calibrated according to the manufacturer's recommended procedure with saturated solutions of known relative humidity (RH). The following freshly prepared calibration standards were used: Lithium Chloride (11% RH), Magnesium Chloride (33% RH), Magnesium Nitrate (53% RH), Sodium Chloride (75% RH), and Barium Chloride (90% RH). To introduce samples to the water activity meter, the dishes were first de-lidded and then swiftly placed into the meter, which was sealed. Readings were taken when the reading was steady (±0.005 water activity unit for 5 min). All determinations were done at 25 • C. Following water activity determinations, the sample dishes were lidded until needed for the uniaxial compression measurements.

Uniaxial Compression Measurements
Uniaxial compression was determined using a LFRA texture analyser (model LFRA1500, Brookfield Viscometers, Harlow, UK). The sample still in the water activity dish was placed on the instrument base plate. The lid was removed from the dish and a 10-mm-diameter flat-ended, acrylic probe was driven at a speed of 5 mm s −1 into the sample. The instrument triggered at 1 g resistance and then compressed the sample to a depth of 2 mm. After compression, the probe was removed at 5 mm s −1 . The force required both to force the probe into the sample and to pull it out where measured. Taking account of the probe contact area, forces were converted to stresses and recorded in Pa.

RESULTS AND DISCUSSION
Lindner and Kinsella [11] made a variety of tahini water mixtures and found them to behave as liquids at water content below 12% and greater than 40%. However, in the range of 14-24% they reported that the rotor of their controlled stress viscometer slipped and in the range of 24-35% the material became inhomogeneous-with the separation of oil. Figure 1 shows the changes in water activity, which we measured in sesame paste with added water. The initial water activity (A w ) of the sesame paste was around 0.58 (±0.01). The gradual addition of water barely raised the A w until about 4% added water was present; further addition led to a gradual rise reaching an A w of 0.90 at about 12%, which corresponds to Lindner and Kinsella's lower limit of viscous behaviour. A subsequent addition of water results in a more gradual rise in the A w until at 28%, where the A w plateaus out at about 0.99 (±0.01). This water content corresponds to Lindner and Kinsella's return to viscous behaviour.
Using uniaxial compression we were able to see how the consistency of the mixtures changes with the addition of water. Figure 2 shows the force needed to push the 10-mm-diameter probe just 2 mm into the surface. The error bars are standard deviations of triplicate determinations. Obviously, at the extremes of hydration (i.e., <12% and >35%) the mixtures behave as liquids and the concept of hardness is inappropriate as the material flows when the probe penetrates the surface, but clearly in the added water range from 12-35% a solid material exists with a peak of hardness around 25%.
In some respects, a single penetration uniaxial compression test is similar to the first cycle of the widely used empirical test protocol "texture profile analysis" (TPA), which is said to mimic what goes on during the first two bites in the mouth during chewing. In terms of TPA terminology, the compressive force in Fig. 1 is akin to "hardness." [12] Removing a flat probe from a sample sandwiched between that probe and a parallel surface is often referred to as a tack test. In TPA terminology this is the "adhesiveness." Figure 3 shows changes in tack as a function of added water. While there is considerable variation between replicate tests, Fig. 3 shows two distinctive peaks of adhesiveness occurring as water is added to sesame paste. It is as though different components of the mixture are separately hydrated. At early stages of hydration the components coalesce and adhere to surfaces on which they contact. It is only when they become fully hydrated that they lose their adhesive capacity. The first of these adhesive peaks occurs at 12-14% added water (when the A w is about 0.90), while the second and higher peak reaches a maximum at 25% added water (with an A w of 0.97). The fact that the A w is less than unity in both adhesive peaks suggests that the substances present have further capacity to bind to water. From a botanical point of view, it makes perfect sense that cell wall and polysaccharide materials should hydrate at a lower A w than enzymes, membranes, and their constituent proteins, thus allowing carbohydrate substrates to be available for metabolism prior to processes, such as germination, when enzymes first become functional.
Whatever the constituents and mechanisms causing these adhesive events to occur, the overall effect of adding water to this suspension of cell debris in oil is to initially raise the viscosity of the solution as the suspended particles start to adsorb water. The mixture gradually hardens reaching a peak that coincides with the maximum adhesiveness at 25% added water. We corroborate Linder and Kinsella's observation of phase separation and liberation of oil at such water contents. Addition of further water results in the formation of an oil in water emulsion (phase inversion) along with softening of the material as the water activity tends towards unity. Water activity has been widely seen as a measure of interaction between water and other components, whereby the water sorption isotherm illustrates the tenacity with which water present is bound to other materials. In the case of these water and sesame paste mixtures, the physical state of mixtures up to 12% water are essentially oil-based suspensions of particles with low water activity. This low A w is presumably partly due to water binding of the cell debris, but also due to the physical isolation of the aqueous phase, separated from the atmosphere by a continuous phase of oil. In this region, further small additions of water become absorbed into the particles, which remain discrete allowing viscosity measurements to be undertaken. [11] With further addition of water, the particles begin to form a network; the adhesive force between particles is apparent at a rheological level with a small increase in tack. Viscosity measurements are no longer possible, but with increasing water addition, solidity develops with the increasing forces needed to compress the mixture and a second peak in the tack occurs. Within this region, there is an expulsion of oil as the inter-particle network becomes more established into a coherent mass.
Finally, at high water levels, the mixture is wholly an oil in water emulsion, which is thinned by addition of further water and which exhibits an A w close on unity, suggesting that all other macromolecules have become saturated with water and excess water is unassociated. In terms of the behaviour of sesame paste in the mouth-the source of hydration is predominantly saliva and the low A w of the paste provides a high water binding capacity able to absorb water from the palate, sucking water from the mouth. In his review of oral food processing, Chen [13] examines the role of saliva in swallowing. He points out that when stimulated the flow of saliva can reach as much as 4.15 cm 3 .min −1 , thus a typical 5 g portion of sesame paste would need to remain in the mouth for about 20 s to acquire enough saliva to lubricate it to 30%, at which point it starts to behave more like a liquid.
Hutchings and Lillford [14] created a model to explain how the bolus develops during mastication. They identified three factors: time in the mouth, degree of structure, and degree of lubrication. They plotted these factors orthogonally (Fig. 4a) and postulated that there was a level of material structure above which we could not swallow. They represented this cut-off as a plane across the "degree of structure" axis, whereby foods that existed above this threshold would need to be masticated until the particle size was small enough to swallow. Similarly they suggested that a minimum level of lubrication was necessary to swallow the bolus and again they represented this threshold as a plane across the "degree of lubrication" axis. The intersection of these two planes formed a solid bar shape (in 3D, the third axis being time) outside of which the bolus needed further oral processing to lubricate or disintegrate, but inside of which the bolus could be cleared. Figure 4b shows two example foods: a liquid with no structure and high levels of lubrication; and "a piece of tough, dry meat which requires extensive mastication and lubrication through the secretion of saliva".
Sesame paste seems to break the rules (Fig. 4c); it starts as a lubricated unstructured liquid but rapidly absorbs water from the saliva, which both dries the mouth and creates a hard sticky mass that is difficult to swallow. [5] Furthermore, as the oil seed paste sticks to the palate and tongue, it becomes inaccessible to the teeth and it is only with time that enough saliva is secreted (or if the subject has a drink), that the now moistened paste can be cleared.

CONCLUSION
Bearing in mind the origin, composition, processing, and structural similarity of sesame paste, peanut butter, cashew nut butter, and other oil seed pastes, it is difficult to dismiss the hard-toswallow behaviour in the mouth as being unrelated. In the case of sesame paste (used in this study), the low initial water activity predisposes the paste to a substantial water adsorption capacity. The resulting sticky mass coats surfaces on which it contacts (such as the palate and the tongue) and it is only with further absorption of water that the stickiness is lost. Further work needs to be undertaken to verify the similarity with other oil seed pastes.