Euhydration depends on adequate Na and water intakes, mediated by shared physiological and neural substrates( Reference De Luca, Vendramini and Pereira 1 , Reference Fitzsimons 2 ). Thirst is an emotion essential for survival, and its physiological instigators are volumetric or osmotic changes in the intracellular, interstitial and intravascular body fluid compartments. Thirst can also be aroused by oral dryness, high concentrations of oral Na, and cognitive, associative and appetitive circumstances( Reference Denton, Shade and Zamarippa 3 ).
However, in the elderly, impaired thirst may compromise euhydration and in extreme heat may significantly contribute to mortality( Reference Cowen, Hodak and Verbalis 4 – Reference Zizza, Kathy and Catherine 9 ). The dangers of hyponatraemia in general are increasingly recognised, and they increase with age. Electrolyte disorders are common among the elderly, often due to the syndrome of inappropriate antidiuretic hormone secretion, or associated with diabetes mellitus and diuretic use in hypertension. Elderly who use both thiazides and benzodiazepines have a more severe degree of hyponatraemia. Even mild electrolyte disorder is associated with mortality and chronic hyponatraemia is associated with bone density loss and fractures( Reference Cowen, Hodak and Verbalis 4 , Reference Fusgen 5 , Reference Liamis, Rodenburg and Hofman 7 , Reference Hoorn and Zietse 10 , Reference Verbalis, Barsony and Sugimura 11 ).
Salt appetite and thirst are intimately related in animals and, as in humans, their neuroendocrine and physiological regulation processes are intimately related too( Reference De Luca, Vendramini and Pereira 1 , Reference Leshem 12 ). Aged rats consume less water and less Na when deprived of either, or stimulated with desoxycorticosterone acetate, and have reduced angiotensin 1 and atrial natriuretic peptide receptor mRNA expression in hypothalamus, which are important central mediators of thirst and salt appetite( Reference Begg, Sinclair and Weisinger 13 – Reference Thunhorst and Johnson 15 ).
There are substantive differences between animals and humans with regard to salt avidity. In fact, it has been proposed that humans do not have a salt appetite as animals do( Reference Leshem 12 ). Indeed, it is not known whether, similar to thirst, salt appetite is compromised in elderly people. If it is, in addition to making a possible contribution to hyponatraemia, it could aggravate nocturia and increase sleep and mood disturbances, which afflict the elderly more and which have been reported to be variously associated with Na intake( Reference Fusgen 5 , Reference Goldstein and Leshem 16 – Reference Vitiello, Prinz and Halter 18 ).
One of the differences between humans and animals is that animals seek Na as any salt, whereas humans only seek NaCl, and therefore herein we shall term the human avidity for salt in its various forms as ‘salt appetite’( Reference Leshem 12 , Reference Leshem, De Luca, Menani and Johnson 19 ).
In the present study, we compared salt appetite in elderly and middle-aged people to determine possible age-related changes. In addition, we compared their 20-year recalled dietary salt intake to determine whether any changes occurred with time or age.
Methods
Participants
Ethics approval for the study was provided by the University of Haifa. A total of sixty individuals from the Arab town of Fureidis volunteered and signed informed consent forms. They were recruited by word of mouth and from day centres for the elderly. Among these volunteers, thirty were middle-aged (49·8 (se 0·7) (45–58) years; sixteen males and fourteen females) and thirty were elderly (73·2 (se 1·1) (65–85) years; fourteen males and sixteen females). Thus, the average difference between the generations was 23·5 (se 1·3) years. Data were collected in 2012–13.
Procedure
The participants were asked to avoid eating and drinking beverages other than water 2 h before the test sessions. There were two sessions, both conducted at the participants’ homes. The first session commenced with signing of the informed consent form including a brief explanation in general terms that the participants would be required to evaluate the taste of various common food items and complete questionnaires about their diet. With the help of the investigators, they completed a brief questionnaire about demographics, smoking status, dieting, health, exercise, nocturia and sleep (Table 1) and completed a dietary questionnaire, the Positive and Negative Affect Schedule (PANAS) mood scale( Reference Watson, Clark and Tellegen 20 ) and salt preference tests.
Table 1 Demographics of the study participants (Mean values with their standard errors, or percentages)
Value was significantly different from that of the 45–58 years age group: * P< 0·05; *** P< 0·001.
Value was significantly different from that of the men: † P< 0·05; †† P< 0·01; ††† P< 0·001.
To estimate salt appetite, the participants were tested for preferred amounts of salt in soup and sugar in tea, followed by a test with oral sprays of NaCl and sucrose solutions. Between the taste tests, the participants were interviewed to complete the dietary, seasoning and preference questionnaire. After finishing the taste tests, while still completing the questionnaire, the participants were invited to eat freely from the salty and sweet snacks provided( Reference Crystal and Bernstein 21 , Reference Kochli, Rakover and Leshem 22 ).
In the second session, a week later, the participants were asked to again complete only the dietary questionnaire, but to recall their diets followed 20 years ago.
The test session for a participant lasted 60–80 min and was carried out by K. H. with one of two trained assistants. Appointments for the test sessions were based on calls to prayer (morning, midday, afternoon and evening). Of the sixty participants, forty were evaluated at the same time of day for both sessions and the remainder at adjacent time periods.
Behavioural tests
A battery of behavioural tests was designed to model different forms of salt intake. Unweighted scores were also combined to provide an operational definition of salt appetite( Reference Leshem 12 , Reference Crystal and Bernstein 21 – Reference Stone and Pangborn 23 ).
Determination of preferred amounts of salt in soup and sugar in tea
For this analysis, tomato soup was prepared by diluting one part of pure, unsalted tomato paste concentrate (22BX) with nine parts of boiled water and tea was prepared with a 3 g tea bag in 1 litre of boiled water. The soup and tea were prepared freshly before each test session and kept in vacuum flasks at approximately 45°C.
The participants were presented with two 200 ml cups of tomato soup, one unsalted and the other with 3·3 % (w/w) NaCl. They were asked to taste the soup in both cups using a 5 ml teaspoon and were provided with a third cup into which the experimenter then poured one-half of the unsalted soup. Then, using the teaspoon, they were asked to add the salted soup to the third cup and taste it until they deemed the mixture most ‘tasty’. The salt concentration of the mixture was determined by weighing the cups, a validated measure of concentration. Preference for sugar in tea was determined similarly, using tea with 20 % (w/w) sucrose( Reference Kochli, Rakover and Leshem 22 ).
Psychophysical ratings of taste solutions in oral sprays using visual analogue scales
A psychophysical rating test was employed to rate intensity and preference for taste solutions. Solutions were diluted to six concentrations with bottled water (9 mg/l Na). NaCl was diluted from 2·56 m by 1/3 steps down to a concentration of 2·5 mm and sucrose from 135 g/l by 1/2 steps down to a concentration of 0·55 g/l. The experimenter sprayed 0·29 ml of each taste solution at specific concentrations in fixed and counterbalanced semi-randomised orders (excluding sequential concentrations) onto the participant's tongue. Using visual analogue scales, the participants rated each concentration for taste intensity (‘how strong is the taste?’ – in Arabic) anchored by ‘don't feel anything’ and ‘very strong’ and for hedonics (‘how tasty is it?’) anchored by ‘bad taste’ and ‘very tasty’.
Quantification of sweet and salty snacks consumed
The participants were invited to eat freely from two familiar commercial salty (890 and 780 mg/100 g Na+) and sweet (120·5 and 146 mg/100 g Na+) snack items presented on separate saucers in unwrapped bite-sized morsels. The number of morsels eaten was recorded discreetly.
Questionnaire
The investigator interviewed each participant using a questionnaire covering sixty-one food items of the common Israeli diet and particular to this community( Reference Kochli, Rakover and Leshem 22 ). The questionnaire provided scores on the following components:
Dietary intakes: The participants were asked about their weekly frequency of consumption of food items and the quantities consumed. These were used to calculate Na+, carbohydrate, sweet carbohydrate, fat and protein contents of their daily diet using nutritional values and portion size tables( Reference Meir and Reshef 24 ).
Seasoning: The participants were asked how much sugar, salt, pepper, and oil/butter they added to season relevant food items (scored on a three-level scale). They were also asked about liking, licking and patterns of ingesting sugar, salt, pepper, and oil/butter. Only seasoning with salt and sugar was analysed.
Salt appetite
‘Salt appetite’ is operationally defined as the (unweighted) mean ranks of the above measures (preferred amount of salt in soup, 640-mm Na spray, number of salty snacks eaten, dietary Na intake and salting). Equivalent sweet preference measures were calculated (preferred amount of sugar in tea, 9 % sucrose spray, number of sweet snacks eaten and sweetening)( Reference Kochli, Rakover and Leshem 22 ). In addition, to examine the relationship of dietary Na intake with the other measures of salt preference and their relationships with other variables without the confound of the largely ‘involuntary’ dietary Na intake, a ‘non-dietary salt appetite’ measure that excluded dietary Na intake was derived.
Positive and Negative Affect Schedule
The PANAS contains two subscales of ten items each, each scored on a five-level scale. Positive affect items reflect a person's enthusiasm, activity, alertness, energy, full concentration and pleasurable engagement. Negative affect (NA) is general subjective distress subsuming a variety of aversive mood states( Reference Watson, Clark and Tellegen 20 ). The PANAS was translated into Arabic and back-translated by different colleagues and corrected accordingly. Two grammatical versions were used, masculine and feminine( Reference Leshem, Sliman and Taweel 25 ).
Statistical analysis
The measures of salt preference, individually scored and also combined into ‘salt appetite’ as described, were compared between the groups using t tests, ANOVA, correlations and regression analysis as appropriate (IBM SPSS Statistics version 19) and are reported.
The α-value was fixed at 0·05, and means with their standard errors are the measures of variability. Non-significant results are not reported unless of specific interest.
Results
Demographics
Middle-aged female participants were found to be more obese than their male counterparts, whereas elderly male and female participants were found to have a similarly increased BMI (Table 1). Despite their greater BMI, middle-aged female participants, as well as elderly male and female participants, reported reduced energy intake (Table 2). Many of the older women reported voluntary restriction of their Na intake. Women reported less exercising than men and did not smoke in this ethnic group.
Table 2 Results† (Mean values with their standard errors)
VAS, visual analogue scale.
* P< 0·05, ** P< 0·01, *** P< 0·001.
† For compactness only significant interactions are reported (see text for lower-order comparisons).
‡ Differences in dietary Na intake are explained by differences in dietary energy intake.
Salt appetite and sweet preference
The 20-year recalled dietary Na intake was greater than the current intake (F(1,56) = 58·6, P< 0·001; Table 2). An interaction of the recalled dietary Na intake with sex (F(1,56) = 5·9, P< 0·05) and with age (F(1,56) = 7·5, P< 0·01) was observed, because it was greater (P< 0·001), especially in women (P< 0·01). Middle-aged participants reported ingesting more dietary Na (P< 0·001) and also greater 20-year recalled dietary Na intake (P< 0·05). However, it is worth noting that the 20-year recalled dietary Na intake reported by elderly participants was similar to the current Na intake reported by the 20 years’ younger group.
The current dietary Na intake adjusted for energy intake (as covariate or as Na density) did not differ significantly by age, sex or their interaction. It was derived from dietary calculations and was correlated with energy intake, both current (r 0·64, P< 0·001) and 20-year recalled (r 0·45, P< 0·001).
Salting habits do not vary with age, sex or recall.
The recalled dietary energy intake was greater than the current intake (F(1,56) = 80·9, P< 0·001). There was an interaction between dietary energy intake and age (F(1,56) = 12·6, P< 0·001), as a greater increase was observed in middle-aged participants, and there was an age effect (F(1,56) = 22·8, P< 0·001), as elderly participants reported reduced intake.
Elderly participants reported greater sweet preference (F(1,56) = 6·3, P< 0·05) and also greater 20-year recalled sweet preference (F(1,56) = 13·0, P< 0·001).
An interaction between taste and sex was observed when assessing the number of salty and sweet snacks consumed (F(1,56) = 9·56, P< 0·005).
An interaction between taste and age was observed when analysing salt appetite and sweet preference (F(1,55) = 9·1, P< 0·005), with the middle-aged participants preferring salt and the elderly participants preferring sweet. Salt appetite was found to be influenced by age (F(1,55) = 4·3, P< 0·05), and a trend for the interaction between age and sex was observed (F(1,55) = 4·0, P= 0·05). Sweet preference was found to not differ by age or sex.
An interaction between sweet preference and age was observed when analysing non-dietary salt appetite (excluding dietary Na intake) (F(1,55) = 4·5, P< 0·05), with a greater sweet preference being observed in elderly participants and a greater non-dietary salt appetite in middle-aged participants. Nevertheless, the effect of age on non-dietary salt appetite was not significant (Table 2). Interestingly, regression analysis showed that non-dietary salt appetite was also correlated with dietary energy intake (t(2,58) = 2·58, P< 0·05), but not with dietary Na intake.
Psychophysical evaluations
In Fig. 1, sensory evaluations of concentration and hedonics for salt and sweet oral sprays are shown. Other than the expected effects of concentration and taste, there were no significant group effects, although women tended to find sucrose more hedonic (men 4·4 (se 0·5); women 5·9 (se 0·5); P< 0·05).
Fig. 1 Taste intensity and hedonics for salt ((a) women and (b) men) and sucrose ((c) women and (d) men) solutions delivered by oral spray. There was no difference in taste hedonics or taste intensity by age or sex. VAS, visual analogue scale. 
, Middle aged (45–58 years), taste hedonics; 
, aged (65–85 years), taste hedonics; 
, middle aged (45–58 years), taste intensity; 
, aged (65–85 years), taste intensity. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn).
Thirst
The means of the appetite measures, including thirst, are given in Table 2. While evaluating the changes in thirst visual analogue scale scores over the duration of the test, an effect of age (F(1,56) = 11·7, P< 0·001) and of sex (F(1,56) = 4·1, P< 0·05) and an interaction between sex and age (F(1,56) = 4·9, P< 0·05) were observed, because elderly participants did not report an increase in thirst, unlike middle-aged participants, and female participants reported being thirstier, particularly elderly women.
Water intake (cups/d) was found to not vary with age, sex or 20-year recall.
Positive and Negative Affect Schedule
Positive affect was lower in elderly participants (27·7 (se 1·4) v. 31·4 (se 0·9), F(1,56) = 5·1, P< 0·05), and NA was significantly higher in women (21·8 (se 1·4) v. 14·2 (se 1·4), F(1,56) = 15·1, P< 0·001), suggesting much higher levels of depression and anxiety than among men.
Hypothesised correlations
Mood
There was no significant relationship between Na and mood (PANAS). However, sweet preference was found to be related to NA in middle-aged female participants (r 0·55, P< 0·05) and to positive affect in the middle-aged male participants (r 0·54, P< 0·05).
NA was found to be correlated with the use of medication in middle-aged participants (r 0·35, P< 0·05).
Nocturia
The night frequency of nocturia was found to be not significantly related to any measure of salt appetite or fluid intake.
Nocturia was found to be correlated with NA (0·29, P< 0·05), and women reported having more frequent nocturia than men (F(1,56) = 12·1, P< 0·001, 2·33 (se 0·23) v. 1·21 (se 0·23)), consistent with their greater NA.
Sleep
Dietary Na intake was found to be related to sleep duration in middle-aged participants (0·40, P< 0·05), but the effect was equally correlated with dietary energy intake (0·41, P< 0·05). Sleep duration was found to be inversely related to sweet preference among women ( − 0·48, P< 0·01).
Elderly participants reported sleeping less at night (4·3 (se 0·3) v. 5·5 (se 0·3) h, P< 0·01).
Discussion
We found no clear impairment of salt appetite in elderly participants when compared with middle-aged participants. Although elderly men (but not women) have a lesser salt appetite (as defined above) than their middle-aged counterparts, this is best understood as an artifact due to reduced energy intake in the elderly, in turn reducing dietary Na intake, and therefore it is not a specific reduction in salt avidity. There were no age-related differences in the intensity of taste or hedonic profile of Na or changes in the reported amount of salting over the years as well as no age-related differences in salting soup or number of salty snacks consumed. No age-related sex differences were observed, although men tended to prefer salt more and women sweet( Reference Leshem 12 ).
Because salt appetite excluding dietary Na intake does not differ with age, it would seem equally unlikely that the reduction in total dietary intake is due to reduced salt appetite in the elderly, despite the close correlation of dietary energy and Na intakes.
In the absence of documentation in this community, we asked the study participants to recall their dietary intake 20 years ago to assess whether diet composition and Na intake differed by generation or had changed during the past 20 years. We found that the recalled dietary Na intake was greater than the current intake, especially in women. However, while elderly participants reported ingesting less dietary Na, their 20-year recalled Na intake was found to match the current intake of the 20 years’ younger group. Moreover, analysis (data not shown) revealed that only two to three dietary items of the sixty-one items differed in the current and recalled diets, although most were present at greater amounts in the recalled diet.
The diet was mostly home prepared. Items eaten most (by energy) were pita and chocolate, followed by cookies, chicken schnitzel and red meat, and then fresh fruit and sweet beverages.
Together, these suggest that the reduced intake in the elderly is indeed due to ageing rather than due to dietary changes over time.
Clearly, diets recalled from 20 years ago are not validated, and a correlation with the current dietary intake may be imposed by the questionnaire and context despite separating the questionnaire administration by days. On the other hand, the interesting correlation with BMI (below) lends some validity to the measure.
In contrast to the intact salt appetite in the elderly, we found that elderly participants were less thirsty at the beginning of the test session and that during the test session they did not report an increase in thirst, whereas middle-aged participants did so by almost 50 %. Elderly participants did not recall greater drinking or liking for water 20 years ago, possibly because impaired thirst may also limit retroactive awareness of thirst.
We found that women were thirstier than men. This confirms an earlier hypothesis that because men concentrate urine more and their thirst/vasopressin system has higher thresholds, they might drink less than women, and this is consistent with findings in older women who drink more than men after exercise( Reference Baker, Munce and Kenney 26 , Reference Perucca, Bouby and Valeix 27 ).
Contrary to our hypothesis, we found no relationship between Na intake and mood measured by the PANAS. Confirming the validity of our measurements, positive affect was lower and NA higher in women in accordance with their known higher levels of depression( Reference Thunhorst and Johnson 15 ). In a sample of some 10 000 people (US National Health and Nutrition Survey, 2007–2008), dietary Na intake was found to be related to depression, while amount of salt preferred was found to be inversely related to depression, suggesting self-medication( Reference Thunhorst and Johnson 15 ). However, the effects are small and may not emerge in a smaller sample, with a different measuring instrument, and ethnically homogeneous. Moreover, none of the present study participants met the criteria for severe depression, so that restricted variability in the PANAS scores may limit the scope for demonstrating any relationship with salt appetite( Reference Thunhorst and Johnson 15 , Reference Aldinger, Stopsack and Barnow 28 ).
Among the middle-aged participants, sweet preference was found to be related to positive mood in men and negative mood in women. Anhedonia to sweet taste is a well-validated model of depression in animal research, but in human research increased intake of sweets is more often associated with depression( Reference Leshem 29 , Reference Jeffery, Linde and Simon 30 , Reference Camilleri, Méjean and Kesse-Guyot 31 ). NA was much higher in women in the present study and may be related to their sweet preference – remarkably similar findings have been reported recently in French women( Reference Camilleri, Méjean and Kesse-Guyot 31 ).
In men, the greater intake of salt, even when adjusted for body weight, and the greater preference for salt might be related to their greater lean body mass and greater sweat rates, possibilities remaining to be investigated( Reference Leshem 12 , Reference Thunhorst and Johnson 15 , Reference Leshem, De Luca, Menani and Johnson 19 ).
Also, in contradiction to our hypotheses, the frequency of nocturia was found to be not significantly related to any measure of salt appetite. Nor was it related to fluid, water, beverage or soft-drink intake (data not shown), did not vary with age, and was not related to sleep duration.
Female participants reported more frequent nocturia when compared with male participants. This has been reported in younger women, but the present study sample differs from that of other studies in which frequency and incidence were found to be higher in men, although, similar to us, they reported similar rates for the elderly( Reference Bosch and Weiss 32 ). It is possible that volume and concentration of night urine are influenced by Na and fluid intakes, which may not influence the frequency of nocturia( Reference Bosch and Weiss 32 ).
The frequency of nocturia is related to NA, possibly reflecting the known relationship of depression and troubled sleep, although we did not find an association of sleep duration with PANAS scores. Sleep duration in this population may be assessed and reported differently – the day is partitioned by calls to prayer (5–6 a day, commencing at approximately 04.00 hours) rather than clock time. In addition, afternoon sleep is customary, especially among elderly women, and may have contributed to the short night-sleep duration reported. Despite this, the known briefer sleep duration in the elderly was clearly evident, suggesting that any robust influence of Na intake or appetite could have emerged. The absence of an association between Na and sleep is in line with an analysis of National Health and Nutrition Survey (USA) data( Reference Thunhorst and Johnson 15 ).
The participants of the present study reported high Na and energy intakes, which were confirmed in a retest of the participants with more extreme values. Both intakes were substantially higher than those from another report from Israel( 33 ). However, the high BMI that we found is generally consistent with that report and more compatible with the high energy intake that we found.
Despite their greater BMI, middle-aged female and elderly male and female participants reported reduced energy intake. Although the under-reporting of dietary intake is known, particularly in obese (e.g. Archer et al. ( Reference Archer, Hand and Blair 34 )), the discrepancy observed in the present study may be explained by prior energy intake, thus the correlation of BMI with the 20-year recalled dietary energy intake among men (r 0·46, P< 0·05), but not with the current energy intake. In addition, we found that more elderly participants reported dieting (data not shown), again suggesting that BMI may have increased earlier in life and is less related to the current energy intake.
It is not clear why salt appetite is not impaired in the elderly, but thirst is. Thirst is aroused by dehydration and depends on physiological substrates that function less well in the elderly possibly related to changes in volumetric, rather than osmotic, thirst mechanisms( Reference Cowen, Hodak and Verbalis 4 – Reference Zizza, Kathy and Catherine 9 , Reference Stachenfeld, DiPietro and Nadel 35 ), although changes in higher cortical functioning might also contribute( Reference Farrell, Zamarripa and Shade 36 ). Although old rats drink lesser amounts of saline solution than middle-aged rats, the amounts of saline ingested are ‘still copious and comprise an unambiguous demonstration of salt appetite in old rats’( Reference Thunhorst, Beltz and Johnson 37 ). Moreover, recent findings in the Fischer 344 × Brown Norway rat suggest that it might be a useful model for mineralofluid regulation in elderly humans because it ‘support(s) the idea that impairments in behaviour contribute more to the waning ability of ageing animals to respond to body fluid challenges than do declines in kidney function. In addition, the results suggest that the behavioural defence of Na homeostasis is less diminished with age compared with other strains so far studied’( Reference Thunhorst, Beltz and Johnson 38 ).
In humans, salt appetite may be a conditioned preference, as we have shown in exercisers losing Na in sweat and in the robust salt appetite persisting over decades among urbanised Bedouin, equalling that of their desert-dwelling encampment contemporaries( Reference Farrell, Zamarripa and Shade 36 , Reference Leshem, Saadi and Alem 39 , Reference Wald and Leshem 40 ). Moreover, early Na loss engenders an enduring enhanced salt appetite in humans and rats( Reference Leshem 12 , Reference Crystal and Bernstein 21 , Reference Leshem 41 – Reference Stein, Cowart and Epstein 43 ). For these reasons, salt appetite in the elderly may be maintained despite physiological changes with age.
Conclusions
We found that salt appetite in the elderly is intact. Salt appetite as operationally defined to represent various forms of salt intake includes questionnaire-derived dietary Na intake, salting, liking for salt, laboratory tests of hedonics of saline solutions, preferred amounts of salt in soup, and number of salty snacks consumed( Reference Leshem 12 ).
Reduced dietary Na intake in elderly men, and the recent nuanced understanding of the lower hypertension risk, cardiovascular health, and increased longevity with Na intake within 2·5–6·0 g/d( Reference Alderman 44 , Reference Strom, Anderson and Ix 45 ), might suggest that Na supplementation in the elderly could be judiciously considered( Reference Fusgen 5 ). Where salt sensitivity is not suspected, salt appetite might be utilised to improve the well-being of the elderly and Na supplementation where hyponatraemia is suspected, and isotonic drinks might be beneficial to counter age-related loss of appetite and to stimulate thirst( Reference Fusgen 5 , Reference Schiffman and Graham 46 ).
Finally, the findings reinforce the view that Na intake in humans is not physiologically driven( Reference Leshem 12 , Reference Leshem, De Luca, Menani and Johnson 19 ), unlike in animals, in which it is exquisitely physiologically regulated, and partially impaired in aged rats( Reference Begg, Sinclair and Weisinger 13 , Reference Thunhorst and Johnson 15 , Reference Thunhorst, Beltz and Johnson 37 ).
Acknowledgements
The present study received financial support from the Salt Institute and the University of Haifa. The funders had no role in the design and analysis of the study or in the writing of this article.
The authors’ contributions are as follows: M. L. formulated the research questions, M. L. and K. H. designed the study, analysed the data and wrote the article; K. H. carried out the experiments.
The authors have no conflicts of interest to declare.
