Table 2 Characteristics of articles eligible for review

Albarracin (2009); Study 1; US [26]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 53 (72% males) [n = 26 in prime condition (assumed)];

Age: 18.90 (1.11); BMI: 22.20 (3.85) (assessment method not reported);

No exclusions

Five print advertisements (duration: not specified)

Experimental: exercise campaign adverts with action words (‘go for a walk’; ‘join a gym’; ‘Go skating’

Control: general adverts with non-exercise messages: ‘make friends;’ ‘be in a group;’ ‘be together’ Attended (participants rated appeal and efficacy of adverts)

No cue validation

Raisins (n = 20) (sweet)

Timing: immediately after cue exposure

Outcome: EI (assessed with piece count)

None

EI was greater in the experimental compared to control condition

No procedures to control for appetite between conditions;

Cover story used. No info about participants’ beliefs about the aims of the study;

Small portion size might have constrained food intake

No test of mechanism

Albarracin (2009); Study 2; US [26]

Laboratory; between-subjects; random assignment to conditions;

Student sample; n = 51 (46% males); [n = 25 in prime condition (assumed)]; Age: 19.26 (1.16); BMI: 22.80 (4.25) (assessment method not reported);

No exclusions

Eight words (duration: not specified)

Experimental: action words (active, go);

Control: neutral words ‘pear;’ ‘moon;’

Subliminal (15 milliseconds)

No cue validation

Raisins (sweet), M&Ms. (sweet, peanuts (savoury) (15 of each)

Timing: immediately after cue exposure

Outcome: EI reported (assessed with piece count)

None

EI was greater in the experimental compared to control condition

No procedures to control for appetite between conditions;

Cover story used; No info about participants’ beliefs about the aims of the study;

Small portion size might have constrained food intake

No test of mechanism

Boland (2013); Study 2^b; US [14]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 149 (46.1% males) in morning [n = 37 in prime and control (assumed)] and afternoon sessions [n = 37 in prime and 38 in control conditions (assumed); Age: 20.17; BMI not reported;

No exclusions

Word search puzzle with thirteen words (duration: not specified)

Experimental: Included seven ‘healthy’ words (energetic, exercise, fitness, healthy, nutritious, strong and thin);

Control: included seven neutral words (alligator, gasoline, magazine, mountain, picture, ranch, shampoo)

Attended (searched for words)

No cue validation

M&Ms. (sweet)

Timing: immediately after cue exposure and while watching a television programme

Outcome: Ounces (weighed intake)

Time of day: morning, afternoon.

No main effect of condition. Significant condition x time of day interaction, food intake was significantly lower in the experimental condition compared to control in the afternoon. Food intake did not differ between conditions in the morning session

Not random allocation to morning or afternoon sessions;

No procedures to control for appetite between conditions (pre-test showed no difference between morning and afternoon, no data reported comparing appetite between experimental and control conditions);

Not clear info about cover story; No participants guessed study purpose;

SD for age not provided, range: 18–27 years. No test of mechanism

Bourn (2015); AU [29]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 99 females (n = 48 in prime condition); Age: 19.35 (2.11); BMI: 23.51 (5.19) (self-reported);

No exclusions

Reality television programme (duration: 10 min)

Experimental: ‘The Biggest Loser’ focused on weight loss and appearance of four female contestants;

Control: ‘The Block’- renovated apartment Attended

No cue validation

Chocolate (sweet), corn chips (savoury) and mixed dried fruit (sweet)

Timing: during cue exposure. Outcome: EI (weighed)

None for food intake

No significant differences in food intake between conditions

No procedures to control for appetite between conditions;

Cover story used;

No info about participants’ beliefs about the aims of the study;

Of the 99 participants, 79 did not consume any test foods.

No test of mechanism

Boyce (2013); NZ [22]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 100 females [n = 50 in prime condition (assumed)]; Age: 21.92 (3.90); BMI: 22.89 (3.18) (measured by researcher);

Exclusions n = 46 (n = 29 excluded for guessing study aims; n = 7 who were obese; n = 7 with food allergies; n = 3 who were middle aged)

Seven magazine advertisements (slideshow) (duration: each slide displayed for 20 s)

Experimental: Five featuring thin and attractive models (product based: (perfume or make up) and two filler adverts .Control: Same as experimental but with thin models removed. Attended (participants were asked to take in as much detail as possible)

No cue validation

Chocolate (sweet) and crispy M&Ms. (sweet)

Timing: after cue exposure participants completed a memory task and an implicit mood task (with positive or negative emotional words) and were then provided with snacks. Outcome: Grams (weighed)

Restraint: continuous (Restraint Scale, Concern for Dieting Subscale only and Dietary Intent Scale) [57]

No significant differences in food intake between conditions. Intake did not differ at varying levels of restraint across conditions

No procedures to control for appetite between conditions however pre-study hunger (7-point scale) was included as a covariate in analyses;

Cover story used and participants who guessed study aims were excluded;

Administering the implicit mood and weight satisfaction task prior to the snack test might have primed positive or negative emotions or negative body image;

Insufficient data available to be included in meta-analysis

Tested implicit mood as a potential mechanism

Boyce (2014); NZ [28]

Laboratory; between-subjects; random assignment to explicit (n = 96) and incidental (n = 78) conditions and then random assignment to prime (explicit n = 47; incidental n = 39) and control conditions;

Student sample, n = 174 females; Age: 20.43 (6.29); BMI: 23.06 (2.75) (measured by researcher);

Exclusions n = 75 (n = 29 for guessing study aims; n = 26 with BMI ≥ 30; n = 20 did not engage with explicit images).

Seven images (duration: two-minute slideshow, each slide shown for 20 s);

Experimental: featured thin women advertising beauty products;

Control: seven neutral images (e.g. furniture)

Attended (told to concentrate on images for a subsequent memory test) and incidental (images were shown in an adjacent room while participants completed a coping style questionnaires).

No cue validation

Pretzels (savoury), savoury crackers, chocolate/peanut M&Ms. (sweet), bite-sized cookies (sweet)

Timing: after completing a lexical decision task and rating hunger

Outcome: Grams (weighed)

Restraint: continuous (Restraint Scale, Concern for Dieting subscale only and Dietary Intent Scale) [57] Level of engagement: explicit versus incidental

Restraint positively correlated with sweet food intake in the explicit experimental condition; No significant associations in the incidental or control conditions; Savoury food intake was lower in the incidental prime condition compared to control. Intake did not differ in the explicit condition or at varying levels of restraint.

No procedures to control for appetite between conditions; post-cue hunger was included as a covariate in savoury food analyses (not sweet);

Cover story used and participants who guessed the study aims were excluded;

In the inadvertent condition, completing a questionnaire on coping styles might have primed confounding constructs;

Insufficient data available to be included in meta-analysis

No test of mechanism

Brunner (2012); Study 1; CH [9]

Laboratory; between-subjects;

Community sample; n = 95 (30% males) (n = 47 in prime condition); Age: 35.4; BMI not reported;

1 outlier excluded (food intake > 3 SD)

Picture on laptop computer screensaver (duration: 5 min)

Experimental: three thin human-like sculptures from Piazza by Giacometti;

Control: Orange and yellow painting by Rothko

Incidental (‘unobtrusively’ displayed on a computer next to participant)

No cue validation

Chocolate pieces (n = 20) (sweet)

Timing: during cue exposure

Outcome: Piece count

None

Food intake was lower in the experimental compared to control condition

Not specified if random allocation to conditions;

No procedures to control for appetite between conditions; hunger (time point collected not reported) was included as a covariate;

Cover story used. No info about participants’ beliefs about the aims of the study;

SD for age not provided, range: 16–74 years. No test of mechanism

Buckland (2013)^c; UK [13]

Laboratory; within-subjects; randomised order of conditions;

Mixed community and student sample; n = 26 females; Age: 30.03 (9.28); BMI: 25.50 (4.56) (measured by researcher);

Exclusions n = 13 (n = 6 excluded for guessing study aims; n = 6 dieting to maintain weight; n = 1 inconsistent lunch times)

Consumption of a preload before evening meal (284 g) (duration: 10 min)

Experimental: salad (100 kcal);

Control: Water

Attended (listed first thoughts associated with preload, reported frequency of consumption and reported a memory associated with salad)

Cue validation: pre-study online survey (n = 230) showed the priming cue (salad) was associated with dieting to lose weight

Evening meal: cheese and tomato pizza

Timing: after cue exposure participants completed appetite and mood ratings, a lexical decision task and were then provided with test meal

Outcome: EI (weighed)

Dieting status: dieting to lose weight, not dieting

Restraint (R) and disinhibited (D) eating styles: HRHD, HRLD, LRHD, LRLD [Three Factor Eating Questionnaire (TFEQ) [58]

Energy intake was lower in experimental condition compared to control. A significant condition x diet status interaction showed non-dieters’ EI did not differ between conditions, dieters consumed significantly less in experimental compared to control

Control procedures used concerning alcohol intake evening before test session, similar physical activity across test days, fast two hours prior to lunch; fixed lunch provided four hours before evening meal;

Cover story used and participants who guessed the study aims were excluded;

Mechanism tested: A lexical decision task assessed goal accessibility (using 15 diet-related words) after cue exposure

Buckland (2014); UK [12]

Laboratory; between-subjects; random assignment to conditions;

Mixed community and student sample, n = 67 females (n = 35 in prime condition);

Age: 23.67 (5.87); BMI: 23.48 (2.88) (measured by researcher);

Exclusions n = 2 (n = 1 food intake outlier; n = 1 BMI outlier)

Nine subliminal images (duration: 23 milliseconds each; two exposure phases)

Experimental: low calorie foods and beverage items (fruits and diet products);

Control: non-food objects

Subliminal Cue validation: participants (n = 55) rated priming cues as being more associated with losing weight compared to control cues

Four snacks – a low and high fat sweet and savoury snack

Timing: after second cue exposure phase participants completed appetite and mood ratings and were then provided with snacks

Outcome: EI (weighed)

Dieting status: dieting to lose weight or maintain weight, not dieting

Restraint and disinhibited eating styles (TFEQ [58])

No main effect of condition. Condition x diet status interaction approached significance; non-dieters’ EI did not differ between conditions; dieters consumed significantly less in experimental condition compared to dieters in control (same results for high restrained high disinhibited eaters)

Control procedures: instructed to fast two hours prior to a fixed lunch and fast between lunch and test session;

Cover story used. Inclusion of participants who guessed the study aims did not affect the results;

Mechanism tested: A lexical decision task assessed goal accessibility (using 4 diet-related words) after cue exposure

Buckland (unpublished); UK [23]

Laboratory; within-subjects; randomised order of conditions;

Mixed community and student sample, n = 30 females; Age: 27.67 (11.54); BMI: 24.92 (4.01) (measured by researcher);

Exclusions n = 16 (n = 14 dieting to maintain weight; n = 2 methodological issues)

Exposure to the sight and smell of an object (duration: 10 min and remained during snack test)

Experimental: a fresh orange

Control: non-food object (soap)

Attended (participants were instructed to intensely smell the cue three times)

Cue validation: pre-study online survey (n = 180) showed the priming cue (orange) was associated with dieting to lose weight

Cheesy bite crackers (savoury), toffee popcorn (sweet), chocolate chip cookies (sweet) and salted crisps (savoury)

Timing: after cue exposure participants completed appetite and mood ratings, a lexical decision task and were then provided with snacks

Outcome: EI (weighed)

Dieting status: dieting to lose weight, not dieting

Restraint (R) and disinhibited (D) eating styles: HRHD, LRHD, LRLD (TFEQ [58])

No main effects and no significant condition x diet status interactions on food intake

Control procedures: Fixed lunch provided two hours prior to test session. Instructed to fast two hours between lunch and test session;

Cover story used. Inclusion of participants who guessed the study aims did not affect the results.;

Mechanism tested: A lexical decision task assessed goal accessibility (using 15 diet-related words) after cue exposure

Harris (2009)^d; Study 2; US [19]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 64 (assumed even split across three conditions) (32% males) [n = 32 in prime condition (assumed)]; Mean age and BMI not reported

Exclusions: n = 4 for guessing study aims

Eleven television commercials during a 16 min comedy television programme (total 11 min including filler commercials)

Experimental: Four featured food and beverages which included a nutrition message (granola bar, orange juice, oatmeal and an instant breakfast beverage;

Control: All non-food commercials

Attended Cue validation: Pre-study (n = 55) showed the priming cue was rated as being more associated with nutrition and health compared to ‘snack indulgent’ commercials (no comparison to control)

Carrots and celery with dip (savoury); mini chocolate chip cookies (sweet) and cheesy snack mix (savoury), trail mix (savoury) and multi-grain tortilla chips (savoury)

Timing: After cue exposure, participants completed a mood assessment and rated hunger and thirst and were then provided with snacks

Outcome: Grams (weighed)

Restraint: low, high (Restraint Scale [57])

Sex: males, females

No significant differences between food intake in the experimental and control condition

No procedures to control for appetite between conditions; hunger at pre- and post-cue exposure was included in analyses;

Cover story used and participants who guessed the study aims were excluded;

Age range: 18–24 years. Means and SDs were obtained from figures

Means and SDs were not available for sex comparisons or restraint (only z-scores)

Tested hunger and mood as potential mediators, no tests of goal priming as a mechanism

Harrison (2006)^e US [32]

Laboratory; between-subjects; random assignment (one of four conditions: control, image only, image with congruent or incongruent text);

Student sample, n = 222 females (control n = 40; image only n = 54; image/congruent text n = 67); Age: 19.62 (1.12); BMI: 22.68 (4.40) (self-reported);

No exclusions

Thirty images displayed on a slideshow (duration: each image shown for 30 s);

Experimental: attractive young women. Three conditions: images alone or images shown with congruent (motivational language to become thinner or more toned) or incongruent text [about Aruba]

Control: no images (one condition)

Attended (participants rated appeal of each image)

Pre-study sample (n = 12) rated priming cues as representative of thin ideals

Full size pretzels (savoury)

Timing: After cue exposure participants completed a questionnaire with self-concept measures and were then provided with snacks

Outcome: Piece count

Self-discrepancy between perceptions of actual body and perceptions of what peers thought they ought to have: low, high (Self-Discrepancy Questionnaire [59])

Intake did not differ between conditions for women with low discrepancy. Women with high discrepancy consumed less in experimental conditions (image and image with congruent text) compared to control (image with incongruent text did not differ to control)

No control procedures for appetite between conditions;

No info about cover story or participants’ beliefs about study aims;

Food intake was assessed in groups of same sex peers (n = 3–9) (risk confounded by social norms); researcher initiated food intake and then recorded participants’ intake (researcher bias);

Three experimental conditions were compared to one control condition; sample sizes for each comparison were adjusted accordingly.

No tests of mechanism

Jansen^f (2002); NL [27]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 36 females (n = 19 in prime condition); Age: 19.8 (1.6)

BMI: 23.01 (3.1) (measured by researcher;

Exclusions n = 4 (n = 3 for guessing study aims; n = 1 food intake outlier)

Eight images on slideshow (duration: each displayed for 15 ms)

Experimental: thin media models;

Control: neutral (office equipment – stapler, pencil, telephone)

Subliminal (15 ms)

Cue validation: pre-study sample (n = 4) rated the priming cues as representing thin ideal body images

High calorie individually chosen snack

Timing: After cue exposure participants rated mood, self-esteem and completed an image awareness task and were then provided with snacks

Outcome: EI (weighed)

Restraint: low, high (Restraint Scale [57])

No main effects of condition and no significant condition x restraint interaction on EI

Control procedures: instructed to eat a small meal and then fast for 2 h prior to test session;

Cover story used and participants who guessed the study aims were excluded;

Administering the self-esteem scale prior to the snack test might have primed thoughts about self-esteem

Details about the specific test foods provided was not available

Mechanisms tested: mood and self-esteem were assessed as mechanisms. No tests for goal priming

Mills (2002) Study 1^g; CA [21]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 73 females (n = 28 in prime condition); Age: 19.72 (1.13); BMI: 23.78 (self-reported; mean of restrained and unrestrained eaters’ BMI)

No exclusions

Twelve laminated magazine adverts (duration: 15 min)

Experimental: Seven with full-body thin and attractive female models and five filler adverts;

Control: all product only adverts

Attended (participants rated adverts on multiple attributes)

Cue validation: pre-study team of researchers rated images as thin and attractive

Three different flavoured cookies (sweet)

Timing: after cue exposure participants completed mood, self-esteem, and body-size perception and were then provided with snacks

Outcome: grams (weighed)

Restraint: low, high (Restraint Scale [57])

Significant condition x restraint interaction: unrestrained eaters’ intake did not differ between conditions, restrained eaters ate more in the experimental compared to control condition

No control procedures for appetite between conditions;

Cover story used, no participants guessed study aims;

Administering the self-esteem scale prior to the snack test might have primed thoughts about self-esteem

BMI SD not available

Mood and body image assessed as potential mechanisms. No tests for goal priming

Minas (2016); US [41]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 161 (51.6% males) (n = 82 in prime condition); Age: 19.9; BMI not reported

No exclusions

Scrambled sentence computer game task (duration: 8 min)

Experimental: included one or two body image words (slim, fit, weight, diet, healthy, slender);

Control: neutral words (worker, room, leaves, bench, dirt, dwell)

Attended (participants created a five word headline from a set of provided words)

No cue validation

Baked or regular Ruffles crisps (savoury)

Timing: After cue exposure participants completed a demographic survey, rated the cue exposure game and were then provided with snacks while watching a television programme

Outcome: EI

(weighed)

Restraint: continuous (Restraint Scale [57])

Sex: males, females

No main effects of condition. Significant condition x sex interaction: males’ intake did not differ between condition; females’ EI was less in experimental compared to control

No control procedures for appetite between conditions;

Cover story used; no info about participants’ beliefs about the study;

SD for age not provided;

Means and SD unavailable for restraint analyses

No tests of mechanism

Papies (2010); NL [16]

Field (butcher’s store); between-subjects;

Community sample, n = 156 (43.6% males) [n = 76 in prime condition; Age: 56 (14.18); BMI: 26.50 (4.45) (self-reported);

No exclusions

Poster on entrance to a Butcher’s store (duration: N/A)

Experimental: Announced a recipe ‘good for a slim figure and low in calories;’

Control: no poster

Incidental (attention was not explicitly directed to the poster)

No cue validation

Meat snacks (e.g. meatballs) (savoury)

Timing: not reported (differed for each participant)

Outcome: Piece count

Restraint: low, high (Restraint Scale, Concern for Dieting subscale only [57])

Sex: males, females

Significant condition x restraint interaction: unrestrained eaters’ intake did not differ between conditions; restrained eaters consumed less in the experimental compared to control condition

Not random assignment to conditions;

No control procedures for appetite between conditions;

No cover story used, given subtle nature of manipulation unlikely participants guessed the aim;

Of the 156 participants, 86 consumed no snacks;

Means and SDs used in meta-analysis were for males’ and females’ intake obtained from author

For subgroup restraint analysis, means, SDs and sample sizes were estimated

No tests of mechanism

Pelaez-Fernandez (2011)^h; CA [18]

Laboratory; between-subjects;

random assignment to conditions;

Student sample, n = 97 females (n = 49 in prime condition); Age: 19.95 (3.75);

BMI not reported;

No exclusions

Magazine covers placed on table while participants completed consent forms (duration: 10 min)

Experimental: fit females/ slim models;

Control: Geographic scenes and furniture

Incidental (attention not diverted to magazines)

No cue validation

Three types of cookies (sweet)

Timing: after cue exposure half of participants completed a goal accessibility task and were then provided with snacks, the other half were provided with snacks immediately after cue exposure and then completed the goal task

Outcome: Grams (author confirmed in correspondence) (weighed)

Restraint: low, high (Restraint Scale [57])

No main effects of condition and no significant condition x restraint interaction on food intake

No control procedures for appetite between conditions;

Cover story used, no info about participants’ beliefs about the study

Mechanism tested: A goal accessibility task (using 8 diet-related words) was completed after cue exposure or after the snack test (order was counterbalanced across participants)

Seddon & Berry (1996); UK [20]

Laboratory; between-subjects; random assignment to conditions;

Mixed community and student sample, n = 74 females (n = 37 in prime condition);

Age: 25.6 (7.7); BMI not reported;

No exclusions

Television commercials (duration: 12 min)

Experimental: featuring thin and attractive females;

Control: neutral, no thin and attractive women

Attended (participants informed they would be asked questions about the adverts after viewing them)

No cue validation

Salted peanuts (savoury), chocolate coated peanuts (sweet), pickled onion savoury snack (savoury)

Timing: After cue exposure, participants completed a self-esteem measure and were then provided with snacks

Outcome: Grams (weighed)

Restraint: low, high (Restraint Scale [57])

Significant condition x restraint interaction; post hoc tests showed no significant differences between conditions at varying levels of restraint (restrained consumed more than unrestrained in prime condition, no differences in control)

Control procedure: two hour fast prior to test session;

Cover story used, no info about participants’ beliefs about study

Administering the self-esteem scale prior to the snack test might have primed thoughts about self-esteem

Mechanism tested: self-esteem assessed

Sellahewa (2015); AU [10]

Laboratory; between-subjects; random assignment to self-control depletion/no depletion and priming/no priming (2 × 2 design);

Student sample, n = 85 (25.9% males) [21 in each except for n = 22 in non-depletion/goal priming (prior to exclusions)] (25.9% males); Age: 20.08 (3.96); BMI: 21.37 (2.65) (self-reported);

Exclusions n = 6 (n = 4 for guessing study aims; n = 2 non-compliant)

Scrambled sentence computer task (duration: determined by participant)

Experimental: nine health-related words (active, exercise, fit, vigorous, healthy, sunscreen, well, wellbeing, wholesome);

Control: nine neutral words (calm, clever, deodorant, generous, logical, practise, rad, sassiness, spiritual) (words were obtained from authors)

Attended (participants rearranged order of sentence to form a grammatically correct sentence)

No cue validation

Chocolates (sweet), savoury biscuits, potato chips (savoury)

Timing: Immediately after cue exposure

Outcome: Grams (weighed)

Levels of self-control: self-control, self-control depletion [achieved by allowing (self-control) or suppressing (self-control depletion) emotional response to humorous video]

Participants consumed significantly less in the experimental compared to control condition. Condition x depletion interaction on food intake was non-significant.

No procedures to control for appetite between conditions; hunger was included as a covariate;

Cover story used and participants who guessed the study aims were excluded;

Only one example of the healthy words were provided.

No test of mechanism

Stampfli (2016); CH [43]

Laboratory; between-subjects; four conditions – low/high cognitive load and priming/no priming (2 × 2 design;

Community sample, n = 128 (26.6% males) [n = 62 in prime condition (obtained by correspondence with author)]; Age: 46.35 (14.20); BMI not reported;

9 excluded for not fulfilling requirements of cognitive load task)

Picture on computer screensaver (duration: approximately 30 s)

Experimental: three thin human-like sculptures by Giacometti moving on a black background;

Control: static white picture

Incidental (screensaver displayed as participants entered room and chose a cubicle and seated themselves)

No cue validation

Crisps (n = 20) (savoury)

Timing: after cue exposure, participants completed a cognitive load task and were then provided with snacks

Outcome: Grams (weighed)

Cognitive load: low (memorise 2 digits), high (memorise 10 digits) cognitive load.

Liking for snack food: low, high snack liking

Food intake was significantly lower in experimental condition compared to control regardless of cognitive load.

The effect of condition was only found in participants with high snack food liking, not those with low snack food liking

Not random allocation to conditions;

No procedures to control for appetite between conditions;

Cover story used, no info about participants’ beliefs about study

No test of mechanism

Stampfli (2017); Study 1; CH [11]

Laboratory; between-subjects; four conditions – healthy/unhealthy food, priming/no priming (2 × 2 design);

Mixed community and student sample, n = 114 (38.1% males) [n = 34 in priming unhealthy food condition; n = 30 in priming healthy food condition; n = 26 in control unhealthy food) (correspondence with author)]; Age: 31.72 (14.11); BMI not reported;

Exclusions n = 19 (n = 18 for guessing study aims and n = 1 for not answering question about study aims)

Picture on laptop computer screensaver (duration: 30 s)

Experimental: three thin human-like sculptures by Giacometti moving on a black background;

Control: no cue (laptop computer closed)

Incidental (screensaver was running while participants entered room and when participants received study instructions (30 s)

No cue validation

20 chocolates (sweet) or 20 blueberries (sweet)

Timing: Immediately after cue exposure

Outcome: Grams (weighed)

Restraint: low, high (German version of the Restraint Scale, Concern for Dieting subscale [60])

Healthiness of food: healthy (blueberries), unhealthy (chocolate)

Food intake was significantly lower in the experimental condition compared to control regardless of snack food healthiness.

The effect of condition was only found in high restrained eaters, not unrestrained eaters.

Not specified if random allocation to conditions;

No procedures to control for appetite between conditions;

Cover story used and participants who had heard about the study before (and thus were aware of the study aims) were excluded;

No test of mechanism

Stein (2016); US [40]

Laboratory; between-subjects; randomised to one of four conditions – self-control/self-control fatigue, priming/no priming (2 × 2 design);

Student sample, n = 84 (34.5% males) (n = 20 in self-control/priming; n = 22 in no self-control/control; n = 21 in self-control fatigue/priming condition); Age: 18.6 (1.0); BMI: 23.03 (3.85) (self-reported);

No exclusions for food intake data (analyses with exercise levels n = 11 excluded for missing data)

Exercise posters (duration: 20 min)

Experimental: Three posters - a man running in the mountains; a photo of the London Olympics with athletes competing in sports; black and white silhouettes of men and women running;

Control: neutral artwork on walls

Attended (researcher delivered script about the posters to ensure participants noticed them)

No cue validation

Cookies (sweet), chocolate (sweet), potato chips (savoury)

Timing: 20 min after initial cue exposure (participants completed a self-control or placebo task during cue and snack test interval)

Outcome: EI (weighed)

Restraint: continuous (Restraint Scale [57])

Self-control: self-control, self-control fatigue [61]

BMI: low, high

Tendency towards compensatory eating in response to physical activity: continuous

Exercise levels: low, high exercisers

No main effect of condition or self-control depletion on food intake.

High exercisers consumed significantly less in experimental compared to control, low exercisers’ intake did not differ between conditions (n = 73).

All other two-way condition x moderator interactions on food intake were non-significant.

No procedures to control for appetite between conditions; hours since last ate was included as a covariate in the analyses;

Cover story used, no info about participants’ beliefs about study;

Age SD computed from SEM;

No test of mechanism

Strahan (2007); Study 1; CA [30]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 26 females [n = 13 in prime condition (assumed)]; Age range 18–21; BMI not reported;

No exclusions

Television commercials (duration: not specified)

Experimental: total six – four contained no people and two featured slim models (Victoria Secret commercial with a supermodel, Dove soap commercial with slim woman);

Control: total four – none contained people (cellular phone, gas station, pharmacy and insurance company)

Attended (participants given goal to remember as much detail as possible about the commercials)

No cue validation

Popcorn, whole-wheat crackers, crackers (Ritz) and pretzels (all savoury)

Timing: Immediately after cue exposure

Outcome: Grams (weighed)

None

Food intake was lower in the experimental condition compared to control

Control procedures: fast three hours prior to test session;

Cover story used, no participants guessed the study aims;

Means for age not available;

No test of mechanism

van Kleef (2011); US [42]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 125 (43.2% males) (n = 67 in prime condition); Age: 20.5 (5.0); BMI: 22.7 (3.3) (self-reported);

Exclusions: n = 3 [n = 2 vegetarians; n = 1 outlier for reported levels of physical activity].

Eight television commercials (duration: see condition descriptions)

Experimental: exercise equipment and services (running shoes, fitness centre, fitness program) (displayed for 4 min 57 s);

Control: did not refer to food or exercise (car insurance, home appliance, pet dog adoption program) (4 min and 59 s)

Attended (participants rated commercials and preferences for favourite commercial)

Cue validation: Priming cues were rated by participants as making them feel more healthy and in shape compared to ratings of control cues

Lunch meal: pasta dish with tomato sauce, salad and chocolate pudding (salad dressing, cheese and drinks were also available but intake was not recorded)

Timing: Immediately after cue exposure

Outcome: EI (weighed)

Restraint: low, high (Restraint Scale [57])

BMI: low, high

Exercise levels: low, high

Exercise intentions: low, high

EI was lower in the experimental condition compared to control.

The effect of condition on EI was only found in participants with a high BMI, participants’ EI with a low BMI did not differ between conditions.All other condition x moderators interactions on EI were non-significant.

No procedures to control for appetite between conditions; time since last ate did not differ between conditions;

Cover story used, no information about participants’ beliefs about study;

No tests of mechanisms

Versluis (2016)ⁱ; Study 2; NL [15]

Laboratory; between-subjects; random assignment to one of four conditions – small/large portion, priming/no priming (2 × 2 design);

Student sample, n = 224 (59% males) [n = 47 in priming small portion condition; n = 55 in priming large condition; n = 66 in control small condition]; Age: 21 (1.6); BMI: 25.61 (5.12) (based on mean of BMI across four cells) (self-reported);

Exclusions n = 34 (n = 19 for guessing study aims; n = 15 due to allergies or diseases)

Four television commercials inserted in a movie clip (duration: two minutes, 30 s)

Experimental: featured ‘healthy’ foods or services (Dannon Light & Fit yoghurt, Weight Watchers, Nike Basketball and Special K breakfast cereal) with messages about resisting tempting foods, dieting and weight loss;

Control: non-diet-related (garden furniture, Intel, Phillips Ambilight, Jeep Renegade, Amazon Kindle, FedEx)

Attended (participants were asked to recall products advertised in commercials)

No cue validation

M&Ms. (sweet)

Timing: during cue exposure

Outcome: Grams (weighed)

Portion size: small (200 g), large (400 g)

Restraint: low, high (TFEQ [58])

Time of day: 9 am–12 pm, 12 pm–5 pm

Sex: males, females

Perceived dieting success: low, high

Hunger and fullness (pre- and post study; 7-point Likert scale)

BMI: low, high

M&M liking: low, high

Consumption frequency of M&Ms.: low, high

Effect of condition on food intake approached significance with lower food intake in experimental compared to control.

The effect of condition on food intake was found only in restrained eaters in the large portion size condition, not small portion size. Unrestrained eaters’ food intake did not significantly differ between experimental and control conditions in small and large portions conditions.

Assumed random assignment based on study 1;

No procedures to control for appetite between conditions (hunger and fullness at pre-cue exposure did not differ between conditions);

Cover story used and participants who guessed the study aims were excluded;

Of the 224 participants, 59 did not consume any test foods

Means and SDs used in meta-analysis were for males’ and females’ intake obtained from author

For subgroup restraint analysis, means, SDs and sample sizes were estimated

No test of mechanism

Werle (2017); Pilot study; FR [44]

Laboratory; between-subjects; random assignment to conditions;

Student sample, n = 95 (48.4% males – based on correspondence with author) (n = 46 in prime condition); Age: 20.2 (0.8); BMI: 21.2 (2.7) (self-reported); 17 excluded (details not specified).

Commercials (duration: 30 s)

Experimental: featured men and women engaged in sports (e.g. rugby, running; Nike);

Control: featured peacocks (Telus telecommunications). Attended (participants rated advert on multiple attributes)

M&Ms. (sweet)

Timing: Immediately after cue exposure

Outcome: grams (weighed)

Sex: males, females

No significant differences in food intake between conditions (effect of condition approached significance, p = .09)

Gender x condition interaction on food intake was non-significant.

No procedures to control for appetite between conditions;

Cover story used, no info about participants’ beliefs about study;

No test of mechanism