Sensory Discrimination Tests for Low- and High-Strength Alcohol

Abstract

Research is limited on consumers’ ability to detect perceptible sensory differences between low- and high-strength alcoholic beverages. This study, therefore, conducted three pilot experiments using ISO sensory analysis methods to assess accuracy for evaluating beverages of different strengths. Participants were food production professionals trained in sensory analysis. Experiment 1 used a wide-range discrimination test to estimate low- to high-strength beverages (0–60% alcohol by volume (ABV) in 10% intervals). Experiment 2 included a narrower range of intermediate to high strengths (25–45% ABV in 5% intervals). Experiment 3 used 3-alternative forced choice tests (ISO 13301) to discriminate between beverages of varying strengths. Experiment 1 (n = 16) indicated that estimation ability was dependent upon the beverages’ ABV; as ABV increased, estimation significantly decreased (p < 0.005). These findings were not replicated in Experiment 2 (n = 13). In Experiment 3 (n = 17), a significant perceptible difference between high- and low-strength samples was observed in two of nine conditions (35% vs. 31% ABV (p = 0.009); 41% vs. 37% ABV (p = 0.037)). While people can detect large differences in beverage ABVs, they may have a moderate to poor ability to discriminate between beverages of similar strengths. These findings provide support for public health interventions that promote lower-strength alcoholic beverages.

1. Introduction

Alcohol consumption, especially heavy episodic drinking, is a major risk factor for disease and injury worldwide [1]. Numerous health policies exist which can reduce alcohol-related harms, the most cost-effective of which are increases in taxation, decreases in availability, and restrictions on marketing [2,3]; however, multiple barriers, including lobbying by industry and associated stakeholders, prevent the implementation of such policies [4,5,6]. Therefore, there is a need to develop population-level interventions to reduce alcohol-related harms that are supported by all stakeholders.

One such intervention would be to reduce levels of the most harmful ingredient in alcoholic beverages, namely ethanol [7,8], via the promotion and incentivization of lower alcohol content beverages [9]. The success of this intervention would depend upon consumers’ ability to discriminate between alcoholic strengths, as well as social factors, which influence alcohol consumption behaviors. That is, if people see no differences between lower- and higher-strength alcoholic beverages, they may be willing to purchase lower-strength beverages. However, if they view higher-strength beverages more favorably, they may be less likely to purchase lower-strength products. The ability of people to discriminate between alcoholic strengths is also relevant to public health campaigns regarding the risks associated with the consumption of mixed drinks, homemade drinks, and other artisanal alcoholic beverages where the alcohol content is unknown and/or where consumers do not always have access to the original product packaging, as is generally the case in restaurants and bars or when served by others. Without the ability to distinguish between lower- and higher-alcohol-content beverages, people may consume larger amounts of alcohol in a shorter period of time.

There is limited research on the ability to distinguish between alcoholic beverage strengths. Studies include investigating the ability to discriminate between typical alcoholic beverages’ strengths against a placebo [10,11], and against alcohol at low doses (e.g., beer: <1% to 3.8% alcohol by volume (ABV) [12,13,14,15,16]; wine: 12.9% ABV [17]). However, these studies are limited by small sample sizes and their designs make them susceptible to biases. Only a few studies have focused on the high strengths of alcohol [18,19,20,21]. The first of these high-strength studies, by Standing and Blackburn [20], had 10 participants who provided their perceptions of alcoholic strengths by magnitude estimations of 10 samples of white rum, with concentrations ranging from 0% to 40% ABV at 4% intervals; participants’ perceptions were shown to be accurate. A more recent study by Standing examined the perception of alcoholic strengths in ‘masked conditions’ by mixing samples with additional tastants (e.g., flavors such as sugar or cranberry) [18]. Standing found that drinkers were inaccurate in their discrimination of alcoholic strengths, resulting in underestimation of the strengths [18]; this observation was also made by Higgs [19].

Due to these mixed findings, there is a need for alcohol discrimination tests with a stronger experimental design to reduce sensory confounds [22]. One approach that addresses these concerns is the use of the International Standardization Organization (ISO) 13301 sensory analysis methodology [23]. ISO 13301 is commonly used in food science to assess people’s ability to detect a perceptible difference between two or more substances or products by measuring detection thresholds. It employs controlled sensory evaluation techniques, based on repeated 3-alternative forced choice (3-AFC) tests, to compare samples that are randomized in presentation to eliminate order bias, sequential discrimination bias, and guessing bias [24]. A similar methodology using the ISO 4120 triangle test [25] was previously used by Lachenmeier and colleagues who identified via nine tests that 36–73% of their sample could perceive taste differences across beverage strengths, with mixed success at intermediate-strength alcoholic beverages [26]. In contrast to the ISO 4120 triangle test, the ISO 13301 3-AFC methodology applied in this study is based on multiple presentations of triple samples at increasing magnitudes, with the magnitude of the different samples being at higher strength.

Given the limited number of studies, and the potential to decrease alcohol-attributable harms via policies that reduce alcohol content, there is a need to determine if individuals can discriminate across a wide range of alcoholic strengths. Therefore, three sensory discrimination experiments were performed in accordance with ISO methodologies to assess (i) how likely participants are to estimate the correct actual beverage strength when presented with a range of low- and high-strength alcoholic beverages, (ii) how much participants’ estimations deviate from the sample’s actual alcoholic strength, and (iii) whether individuals detect a difference between a higher-strength alcoholic beverage and two identical lower-strength beverages.

2. Materials and Methods

A series of three experiments were conducted to explore participants’ ability to perceive low- and high-alcoholic-strength beverages by taste. Experiments 1 and 2 were based on the principles of ISO 11056 magnitude estimation methodology [27]; however, the participants were constrained to using specific ABV responses in a predetermined range. Second, the values assigned to the samples were not ratio-based. Experiment 1 assessed broad estimation accuracy with 10% ABV increments between 0% and 60% ABV. Experiment 2 assessed more precise estimation accuracy with 5% ABV increments between 25% and 45% ABV. Experiment 3 used ISO 13301 3-AFC methodology to detect if participants could detect differences between similar strength samples with 4% ABV differences. Acceptability of the beverage samples was not measured in these experiments.

2.1. Participants

The recruited participants were all professionals in food production enrolled in a state-level food inspector trainee program. The program consisted of a 6-month curriculum of theoretical and practical training including standard sensory analysis methodologies, without specific emphasis on alcohol strength discrimination. The experiments were conducted during an internship of the participants at the CVUA Karlsruhe, Germany.

2.2. Study Materials

All alcoholic drinks were obtained from local supermarkets in Karlsruhe, Germany. The adjustment of alcoholic strength was conducted with food-grade nondenatured neutral alcohol of agricultural origin (95% ABV) and with distilled water according to Lachenmeier et al. [26]. The testing solutions were prepared by first diluting commercial vodka (Absolut vodka, Pernod Ricard Deutschland GmbH, Cologne, Germany) or white rum (Bacardi Carta Blanca, Bacardi GmbH, Hamburg, Germany) to the mean concentrations. In a subsequent step, the average solution was adjusted to a lower strength by adding water or to a higher strength by adding neutral alcohol. This two-step process was designed to ensure that all testing solutions contained the exact same amounts of flavor compounds from the original vodka or white rum. This approach prevented differentiation based on flavor differences rather than ethanol content. Vodka and white rum were selected because they have minimal taste components (e.g., impurities from fermentation), allowing for a clearer representation of ethanol’s influence.

All samples (10 mL each) were provided to participants in cups. Other materials provided to participants included a bottle of drinking water, a rinsing cup, a study questionnaire, and a pen.

2.3. Protocol

2.3.1. Experiments 1 and 2

In Experiment 1, participants were provided seven samples of vodka and a reference sample with 30% ABV to serve as a standardized anchor to aid participants in calibrating their estimations of the samples’ alcohol strengths. Participants were instructed to taste each sample without swallowing and then guess its strength from 0% to 60% ABV in 10% increments. In addition to these test samples, participants were given a reference sample, which they could re-taste as needed and told the strength of this reference sample (30% ABV). If they were unsure of the sample’s alcoholic strength, they were instructed to guess at random. There were no duplicate samples, and the order of the sample presentation was randomized. Participants were also instructed to take a break and cleanse their palates between each sample testing. Experiment 2 followed the same test protocol, but with five samples of white rum between 25% and 45% ABV in 5% increments to assess estimation accuracy with the narrower band of alcoholic strengths. The reference sample had an ABV of 35%. For experiment 1, participants were informed about the possible ABV values for the alcohol samples they were given. For experiment 2, participants were not told the possible ABV values of the alcohol samples they were given.

2.3.2. Experiment 3

A 3-AFC test was conducted over six conditions. Per condition, each participant received three vodka samples, each labeled with a random three-digit testing number. Two samples had identical alcoholic strengths, and one sample had a 4% higher strength (tested ABV levels: 35%, 37%, 39%, 41%, 43%, 45%, 47%, and 49%). The higher-strength sample was randomly assigned to one of the three positions (left, center, or right). Participants were instructed to taste the samples without swallowing and select the sample with the higher alcohol content. If participants could not detect any differences between samples, they were instructed to select at random. Breaks between sample testing followed the same process as in Experiments 1 and 2.

2.4. Statistical Analysis

2.4.1. Experiments 1 and 2

For experiments 1 and 2, statistics on the mean and the standard deviation of this mean were estimated. To account for the repeated measures design, repeated measures logistic regression models were used to estimate the odds of a correct guess of alcoholic strength as a fixed effect given the actual strength of the samples. To account for the clustering of guesses, participant ID was accounted for as a random effect. Along with odd ratios (ORs), 95% confidence intervals (CIs) were reported.

Linear mixed models (LMMs) were used to estimate the absolute differences between participants’ guesses and the actual strength of the alcohol samples. The fixed effect predictor variable was the strength of the sample. To account for clustering in the absolute differences in participants’ alcoholic strength estimations, participant ID was included as a random effect using a random intercept covariance structure. Exact binomial tests were utilized to test for the significance of the ability of participants to estimate individual alcoholic beverage strengths.

2.4.2. Experiment 3

Individual best-estimate thresholds (BETs) were estimated based on the geometric mean of the last incorrectly assessed concentration and the next higher concentration for which a correct response was given. If the participant failed to identify the test sample at the highest presented concentration, the hypothetically next higher concentration was used to estimate the BET. If the participant identified the correct test sample from the first presented concentration, the hypothetically next lower concentration was used to estimate the BET.

Exact binomial tests were utilized to detect the minimum number of correct responses to conclude that a perceptible difference exists between alcoholic strengths with an α-level of 0.05 and to test for the significance of the ability of participants to discriminate between alcoholic beverages. The exact binomial tests utilized a null probability of 1/3 (i.e., the probability of correctly grouping alcoholic beverages by chance).

A secondary analysis was performed for Experiment 3 whereby we estimated the power to detect a significant difference in the ability of participants to discriminate between alcoholic beverages for each test (based on the point estimates of the proportion of correct assessments) and the sample size required to achieve a power of 0.8 (based on the point estimates of the proportion of correct assessments), assuming that 10%, 20%, 30% and 40% of participants could accurately discriminate between beverages with differing ABVs accounting for chance.

All analyses were completed in R software (version 4.2.2) using package ‘lme4’ [28,29].

In all three experiments, we estimated the proportion of the sample who can accurately discriminate alcoholic beverages by ABV levels. This proportion accounts for the probability of identifying an alcoholic beverage’s ABV by chance and was estimated according to Formula 1. The probability of guessing the sample by chance was 1 in 6 for Experiment 1, approximately 1 in 5 for Experiment 2, and 1 in 3 for Experiment 3.

$$\begin{gathered} ProportionofthesamplewhocanaccuratelyidentifyABVlevelsaccountingforchance \\ ={\displaystyle \frac{(ProportionwhoidentifiedbeverageABV-ProbabilityofidentifyingABVbychance)}{(1-ProbabilityofidentifyingABVbychance)}} \end{gathered}$$

(1)

2.5. Ethics

Ethical review and approval were waived for this study because the affiliated institutes do not require ethical clearance for sensory analysis of foods [30]. Nevertheless, the authors considered guidelines for ethical and professional practices for the sensory analysis of foods. Potential adverse effects for the assessors were excluded as only small amounts of spirits were taken into the mouth; these amounts were considerably below toxic levels, even assuming that an assessor might ingest the portion provided and not spit it out according to typical practices of testing spirits. All assessors had received training in the sensory analysis of foods. All assessors consented to participation in the study.

3. Results

3.1. Experiment 1

A total of 6.2% (n = 1), 6.2% (n = 1), 25.0% (n = 4), 37.5% (n = 5), and 25.0% (n = 4) of participants correctly assessed one, two, four, five, and seven vodka samples, respectively (see Figure A1 in the Appendix A). The LMM had an R² of 63.9% and indicated that participants were able to significantly assess the ABV of the alcoholic beverages compared to what would be expected by chance (p < 0.001). A moderate level of between-subject clustering was observed (Intraclass Correlation Coefficients (ICC) = 0.678; 95% CI: 0.443, 0.914).

At the sample level, participants were significantly able to correctly assess all beverage strengths when compared to chance (see

Table 1. Estimates of participants’ accuracy for estimating alcoholic strengths in Experiments 1 and 2 based on magnitude estimation methodology.

Experiment	Sample ABV (%)	Assessment of Sample ABV (%)		Correct Assessments (n)	Proportion of Correct Assessments (%)		Sample Specific p-Value	Proportion of the Sample Who Can Accurately Discriminate *		Experiment p-Value
		Mean	Standard Deviation		Point Estimate	95% Confidence Interval		Point Estimate	95% Confidence Interval
Experiment 1, vodka (n = 16)	0	3.1	12.5	15	93.8	69.8–99.8	<0.001	92.7	64.7–99.8	<0.001
	10	12.5	12.9	14	87.5	61.7–98.4	<0.001	85.4	55.3–98.2
	20	25.0	11.0	12	75.0	47.6–92.7	<0.001	70.8	38.9–91.5
	30	30.0	9.7	8	50.0	24.7–75.3	<0.001	41.7	12.1–71.2
	40	42.5	5.8	10	62.5	35.4–84.8	<0.001	56.3	24.7–82.3
	50	40.6	14.4	6	37.5	15.2–64.6	0.019	27.1	1.1–58.7
	60	53.8	15.4	12	75.0	47.6–92.7	<0.001	70.8	38.9–91.5
Experiment 2, white rum (n = 13)	25	30.5	7.2	5	38.5	13.9–68.4	0.154	23.1	0.0–60.5	<0.001
	30	31.6	7.8	3	23.1	5.0–53.8	0.732	3.8	0.0–42.3
	35	34.2	5.5	2	15.4	1.9–45.4	1.000	0.0	0.0–31.8
	40	33.8	8.2	3	23.1	5.0–53.8	0.732	3.8	0.0–42.3
	45	41.3	5.5	8	61.5	31.6–86.1	0.001	51.9	14.5–82.7

* These proportions account for the chance discrimination rate (i.e., the number of people expected to guess correctly by chance).

). Participants overestimated the ABV of four of the seven samples (0%, 10%, 20%, and 40% ABV) and underestimated the ABV of two of the seven samples (50%, and 60% ABV). The LMM indicated that the ability to estimate the strength of an alcoholic beverage was affected by the ABV of the beverage assessed. On average, for a 10% increase in the sample ABV, the odds of accurately assessing the beverage’s ABV decreased by 30% (odds ratio (OR): 0.70; 95% CI: 0.55, 0.90; p-value = 0.005).

3.2. Experiment 2

A total of 23.1% (n = 3), 38.5% (n = 5), 15.4.0% (n = 2), 7.7% (n = 1), 7.7% (n = 1), and 7.7% (n = 1) of participants correctly assessed zero, one, two, three, four, and five white rum samples respectively (see Figure A2 in the Appendix A). The LMM had an R² of 19.0% and indicated that participants were able to significantly assess the ABV of the alcoholic beverages compared to what would be expected by chance (p < 0.001). No between-subject clustering was observed (Intraclass Correlation Coefficients (ICC) = −0.044; 95% CI: −0.068, 0.122).

At the sample level, participants were significantly able to correctly assess all 45% ABV beverages. The difference in the ability of participants to accurately discern beverages with ABV at 25%, 30%, 35%, and 40% compared to chance was not statistically significant (i.e., p > 0.05). Participants overestimated the ABV of two of the five samples (25% and 30% ABV) and underestimated the ABV of three of the five samples (35%, 40%, and 45% ABV). The LMM indicated that the ability to estimate the strength of an alcoholic beverage was not significantly affected by the ABV of the beverage assessed.

3.3. Experiment 3

The series of ISO 13301 3-AFC tests indicated that a significant perceptible difference was observed in two of the six vodka conditions (31% vs. 35% ABV; 37% vs. 41% ABV) (see

Table 2. Results of International Standardization Organization 13301 3-alternative forced choice sensory analysis tests for vodka samples (n = 17).

Sample ABVs (%)	Correct Guesses (n)	Proportion of Correct Assessments (%)			Proportion of the Sample Who Can Accurately Identify the ABV of the Sample Accounting for Chance *
		Point Estimate	95% Confidence Interval	p-Value	Point Estimate	95% Confidence Interval
31/35	11	64.7	38.3–85.8	0.009	47.1	7.5–78.7
33/37	9	52.9	27.8–77.0	0.120	29.4	0.0–65.5
35/39	6	35.3	14.2–61.7	1.000	3.0	0.0–42.5
37/41	10	58.8	32.9–81.6	0.037	38.2	0.0–72.3
39/43	9	52.9	27.8–77.0	0.120	29.4	0.0–65.5
41/45	7	41.2	18.4–67.1	0.607	11.8	0.0–50.6

* These proportions represent the proportion of people who can accurately discriminate ABV, accounting for the chance discrimination rate (1 out of 3 people will be expected to guess correctly by chance).

). However, it should be noted that in only one of the six conditions (31% vs. 35% ABV) the 95% CIs of the proportion of the sample who accurately identified the ABV of the sample, accounting for chance, included the null (i.e., 0% of people were able to accurately identify between alcoholic beverages by ABV, accounting for chance). For the 31% vs. 35% ABV comparison, 64.7% (n = 11) of participants correctly discriminated the 35% ABV sample, and for the 37% vs. 41% ABV comparison, 58.8% (n = 10) of participants correctly discriminated the 41% ABV sample. For the non-significant differences which were observed, 52.9% (n = 9), 35.2% (n = 6), 52.9% (n = 9), and 41.2% (n = 7) were able to correctly discriminate the higher-strength sample for the tests which compared beverages with ABVs of 33% and 37%, 35% and 39%, 39% and 43%, and 41% and 45%. The probability of correctly discriminating the higher-strength sample from the lower-strength samples in the ISO 13301 3-AFC tests is 1 in 3. Overall, the BET for ABV was 42.8% with a standard deviation of 2.7%.

The power to detect significant differences in the ability of participants to correctly discriminate between beverages with differing ABVs varied from 0.053 to 0.749 (see Appendix A 

Table A1. Results of International Standardization Organization 13301 3-alternative forced choice sensory analysis tests for vodka samples (n = 17).

Sample ABVs (%)	Correct Guesses (n)	Estimated Proportion of Correct Assessments (%)	Power (under Current Sample Size)	Sample Size Required to Achieve a Power of 0.8	Sample Size Required to Achieve a Power of 0.9
31/35	11	64.7	0.75	20	26
33/37	9	52.9	0.38	50	67
35/39	6	35.3	0.05	4601	6159
37/41	10	58.8	0.57	30	40
39/43	9	52.9	0.38	50	67
41/45	7	41.2	0.10	298	399

). Furthermore, to achieve a power of 0.8, assuming 10%, 20%, 30%, and 40% of participants are able to accurately discriminate between beverages with differing ABVs, accounting for chance, (this corresponds to 40.0%, 46.7%, 53.3%, and 60.0% correctly identifying the ABV), a sample size of 410, 106, 48, and 27 would be required, respectively.

4. Discussion

The present study aimed to assess participants’ accuracy in estimating or discriminating between alcoholic beverages of differing strengths using two magnitude estimation experiments and one ISO 13301 3-AFC Sensory Analysis experiment. The findings revealed mixed accuracy. In Experiment 1, where absolute differences of 10% ABV were assessed, participants were significantly able to estimate the ABVs of different beverages; however, as ABV increased, participants’ accuracy in estimating beverage ABVs decreased. Experiment 2 showed significant accuracy for estimating between alcoholic strengths of 5% ABV; however, participants exhibited less ability to estimate the strengths of alcoholic beverages with a 5% ABV difference compared to those with a 10% ABV difference. Furthermore, Experiment 2 showed there was no effect of the level of ABV on participants’ accuracy. Experiment 3 observed a significant perceptible difference between higher- and lower-strength samples in only two out of nine conditions; however, this lack of significant findings may be due to a lack of power to detect significant differences (due to the small sample size, n = 17). Collectively, these preliminary experiments indicate that participants exhibit strong accuracy in perceiving beverages with 10% absolute differences in ABV compared to 5% ABV, and limited discrimination perception for 4% absolute differences in ABV.

Our findings align with previous laboratory studies [18,19,20], particularly Lachenmeier and colleagues’ alcohol discrimination ISO 4120 triangle test experiment [26]. In their study, participants were able to distinguish between vodka samples with up to a 10% strength difference, with an upper threshold above 40% ABV. Our study corroborates these findings, demonstrating a decreasing trend in discrimination accuracy with 10% ABV differences and an upper threshold of almost 43% ABV in the 3-AFC test. This trend is consistent with the Just Noticeable Difference principle [31], which suggests that as the intensity of the stimulus increases, a greater difference is required for accurate discrimination. This decrease in accuracy may also be attributed to the intense burning taste of ethanol in high-strength beverages [32], such as spirits and overproof alcohol, which likely starts to mask discrimination sensitivity near the 40% ABV threshold. This is also consistent with Nolden and Hayes [32], who observed that as ethanol concentrations increased, the intensity of burning and tingling sensations increased at a greater rate. This may therefore mask the ability to detect or quantify small differences at higher alcohol strengths.

Consequently, these preliminary results provide support for promoting lower-strength spirits as individuals may have limited ability to detect lower ethanol concentrations. These findings also are useful in the formulation of future studies. In particular, the results of this study (point estimates of the estimated discrimination between alcoholic beverages) may be used, in part, to estimate the required sample sizes for future studies which intend to examine in greater depth the ability of people to discriminate between beverages with differing ABVs.

The findings of this study are limited by numerous factors. First, the sample sizes were small, and therefore only large differences in the assessment of participants’ ability to detect differences between alcoholic beverages could be statistically detected. Second, the participants of this study were professionals in food production, enrolled in a state-level food inspector trainee program, and not representative of the general alcohol consumer. Comparatively, untrained consumers would be hypothesized to experience at least the same level of difficulty, or greater difficulty in distinguishing between alcoholic beverages. Third, data about the participants were not collected and therefore we cannot assess if any demographic factors affected discrimination ability. Previous studies have shown that food perception ability may vary based on genetic factors, and this ability may be impaired by damage to the taste system (e.g., nerve damage) as well as by smoking [33,34,35]. Furthermore, as individual characteristics of participants were not collected, it is not possible to assess the applicability of the results to the general population. Fourth, our study only examined alcoholic strength perception in a sober state and did not assess how sensory perception may change with elevated blood alcohol concentrations, thereby restricting the generalizability of our findings as many drinking occasions involve varying degrees of sensory impairment due to intoxication. Fifth, Experiments 1 and 2 did not include an assessment of whether participants could differentiate samples even if their estimations were incorrect. Lastly, our study examined the effects of participants’ ability to detect differences between alcoholic beverages using only white rum or vodka mixed with water. Therefore, the results of this study do not extend to beer, wine, or mixed drinks where the presence of multiple tastants may alter an individual’s ability to detect or mask differences in alcoholic beverage strength and taste perceptions. This can be due to factors such as mixture suppression as beverage complexity increases, [36] ethanol’s influence on complex flavor perceptions [37], and the potential alteration of flavor profiles due to the dealcoholization process if the strength is lowered for beer or wine [35].

In Experiment 3, there are conflicting findings between the 95% CIs and significant tests regarding whether people can discriminate between beverages with an ABV of 37% vs. 41%. In some instances, 95% CIs can include the null value (in this case, 0% of the sample was able to accurately identify the ABV accounting for chance), even when the statistical test shows a significant proportion of the sample accurately identified the ABV, accounting for chance. This difference in findings is due to the distinct methodologies used to construct 95% CIs and perform significance tests [38]. The 95% CIs represent a range of means assuming the sample is representative of the true effect. In contrast, a p-value represents the probability of observing the current sample if, in truth, there is no effect (i.e., no one can accurately identify the ABV of alcoholic beverages, accounting for chance).

The findings of the three experiments may offer implications for public health interventions aimed at reducing alcohol-attributable harms. As the alcohol content increased, participants’ perception abilities declined, resulting in a higher frequency of both overestimation and underestimation of the actual alcohol content. This highlights the potential effectiveness of promoting lower-strength beverages as an intervention strategy. Notably, individuals are less likely to perceive a reduction in alcohol content among higher-strength beverages, thereby supporting the implementation of policies advocating for lower-strength alcohol products. Interestingly, there is industry support for such a policy change. Anheuser-Busch InBev, a major global beer producer, has already launched an initiative encouraging consumers to transition to lower-strength beers, with a target of 20% of their total beer volume consisting of low-strength options by 2025 [39].

Additional support for this public health intervention can be found in non-laboratory studies. For example, a randomized controlled trial that manipulated the ethanol content of alcohol served at university fraternity parties found that low-alcohol beer was not recognized by drinkers and did not lead to increased drinking volumes. Interestingly, the titration hypothesis, namely that people will consume more lower-alcohol-content beer in order to become intoxicated, was not supported [40]. Furthermore, studies conducted in Australia demonstrated that a rapid increase in consumption of low-strength beer in the Northern Territories was accompanied by a reduction in acute and chronic mortality [41,42,43].

5. Conclusions

The three experiments highlight the limitations of individuals to accurately estimate or discriminate between different alcoholic beverage strengths for absolute ABV differences of 4% to 5%. Future studies should aim to replicate and extend these preliminary findings in the general population with a large sample of alcohol consumers via randomized controlled experiments involving diverse beverage types, acceptability measures, and varying levels of intoxication. If the findings are consistent, this may guide the development of interventions aimed at promoting and incentivizing the consumption of low-strength alcoholic beverages.