Introduction
Artificial eyes, ears, and noses for stronger, safer troops A layer of mucus dissolves the arriving scents and separates out different odors molecules so that they arrive at the receptors at different speeds and times. The brain is able to interpret this pattern to distinguish a diverse range of smells.
In contrast, an artificial nose consists of a much smaller array of chemical sensors, typically between six and 12, connected to a computer or neural network capable of recognizing patterns of molecules.
A neural network is a collection of computer processors that function in a similar way to a simple animal brain. The nose doesn’t have a specific receptor for the smell of roses; instead it detects a particular mixture of sweet, sour, and floral, which the brain recognizes as a rose. Similarly, the Tufts artificial nose has 16 fluorescent sensor strips, each sensitive to a different range of molecules, and a computer that interprets their response pattern to determine whether or not they have sniffed a mine. While this method can be better at filtering out false alarms than the Fido approach, it may not be quite as sensitive to explosives-related chemicals.The human nose contains more than 100 million receptors.
Initially developed as laboratory instruments, electronic noses that mimic the human sense of smell are moving into food, beverage, medical, and environmental applications. The Researchers and manufacturers alike have long envisioned creating devices that can ’smell’ odors in many different applications. Thanks to recent advances in organic chemistry, sensor technology, electronics, and artificial intelligence, the measurement and characterization of aromas by electronic noses (or e-noses) is on the verge of becoming a commercial reality.
Odours as Affective-processing Context for Word Evaluation: A Case of Cross-modal Affective Priming
Dirk Hermans, Frank Baeyens, and Paul Eelen
Department of Psychology, University of Leuven, Belgium
In a recent series of priming studies (e.g. Hermans, De Houwer, & Eelen,
1994), it has been demonstrated that response latencies to affec tively
valenced targe t stimuli are mediated by the affe ctive relation between the
valence of the target and the valence of the priming stimulus that immediately
precedes the target. If prime and target share the same valence (e.g.
positive-positive), response latencies are facilitated as compared to trials for
which prime and target are of opposite valence (e.g. negative-positive). This
line of research provides strong support for the assumption that humans
continuously evaluate exte rnal stimuli in an automatic fashion, which is
one of the central premises in a number of recent cognitive-representational
models of emotion. Whereas in previous affe ctive priming studies only
visual stimuli (words, simple line drawings, pictures) have been used as
primes and targets, in the present experiment, positive and negative odours
were used as primes, and words as targets. Results showed that target words
were evaluated faste r if preceded by a similarly valenced odour, as compared
to affe ctively incongruent odour-word pairs. This effect was restricted to the
female subjects, a fac t which is attributed to general gender differences in
odour perception.
INTRODUCTION
In a recent series of priming experiments (e.g. Bargh, Chaike n, Govender,
& Pratto, 1992 ; Fazio, Sanbonmats u, Powell, & Kardes, 1986; Hermans,
De Houwer, & Eelen, 1994, 1996 ), it has been demonstrated repeatedly that
COGNITION AND EMOTION, 1998, 12 (4), 601± 613
Requests for reprints should be sent to Dr Dirk Hermans, Department of Psychology, University of Leuven, Tiensestraat, 102, B-3000 Leuven, Belgium; e-mail: Dirk.Hermans@ PSY.KULEUVEN.AC.BE The authors would like to thank Rika Baert, Helma Van Den Bergh, Nico Nootebaert (University of Amsterdam), Marcel Lenaerts, Rik Delabastita, Noel Bovens, the Redactie raad Gedragstherapie, and the Peperkoeken Huizeke for their logistic and technical support. The ® rst authors are Post-doctoral researchers (Fund for Scienti® c Research, Flanders, Belgium).
q 1998 Psychology Press Ltd
response latencies towards affectively valenc ed stimuli are mediated by the
affective context against which they are presented. The basic paradigm of
these studie s, which has become known as the affective priming paradigm,
is in fact a modi® ed version of the traditional sequential priming procedure (Neely, 1991). In a standard affective priming study, a series of target words (e.g. construc tive, jealous ) is presented, which have to be evaluate d as quic kly as possible as either ``positive ’ ’ or ``negative ’ ’ . Each target word is preceded by a prime stimulus , which can be either positive , negative , or neutral (e.g . music, dentist, circle), and which has to be ignored by the subject. Of crucial importance in these priming studie s is the affective relation between the valence of the prime and the valence of the target, which is typic ally manipulate d over three levels. Prime-target pairs can be either affectively congrue nt (e.g. music-constructive; dentist-je alous), affective ly incongruent (e.g. dentist-constructive; music-jealous ), or affectively unrelated (e.g. circle-construc tive; circle-jealous ; control pairs). It has now repeatedly been demonstrated that the time needed to respond to the targe t stimuli is mediated by this affective relation. Response latencies are shortened on affectively congrue nt prime-target trials , as compared to the control trials, and are relative ly inhibite d for affectively incongruent trials. This data pattern can only be explaine d if one assume s that subje cts evaluate the primes, even though they are asked to ignore these stimuli. In most affective priming studie s, the interval between the onset of the prime and the onset of the target, or the Stimulus Onset Asynchrony (SOA), was only 300msec (Prime = 200msec; Inter Stimulus Interval; ISI = 100msec), which is assumed to be too brief for subje cts to deploy controlled response strategies (Neely, 1977). Following this line of argument, the affective priming effects, which are observed under these conditions, should be attributed to automatic processes. Moreover, if the SOA was prolonge d from 300msec to 450msec or 1000msec (Fazio et al., 1986; Hermans, 1996; Hermans et al., 1994), the priming effect disappe ared. This is an important ® nding, as it is an indire ct but strong indication that automatic processes are responsible for the effect. Indeed, if affective priming should be attributed to controlled proc esses, one would expect even stronge r or at least similar results if subje cts are provided more time to process the prime-target relation. Moreover, the affective priming effect could be replic ated, using even shorte r intervals (SOA 150; SOA 0) (Hermans, 1996; Klaue r, Rossnagel, & Musch, 1997 ). Nevertheless, these SOA effects provide only indire ct evide nce for the automaticity of these priming effects. Probably the strongest indication for the automatic nature of the affective priming effect stems from research by Greenwald (Greenwald, Klinge r, & Liu, 1989; Greenwald, Klinge r, & Schuh, 1995), who reliably demonstrated affective priming with subliminally presented primes.
602 HERMANS, BAEYENS, EELEN
Following the original studie s (Fazio et al., 1986), the generality of this
affective priming effect has now been demonstrated using different types
of response tasks such as pronunciation (Bargh, Chaiken, Raymond, &
Hymes, 1996; Hermans et al., 1994) and lexical de cisions (Hill & Kemp-
Wheeler, 1989 ; Wentura, 1996 ). Also, several aspects of the original procedure (Fazio et al., 1986), have been shown not to be necessary preconditions for the effect to occur (Bargh et al., 1992; Hermans et al., 1994 ). Nevertheless, in some studie s the basic affective priming effect could not be replicated. Greenwald et al. (1995 ), for example , obtained no priming effects with visible prime s, and SOAs of 250msec to 300msec. Also in our own research (Hermans, 1996 ), the basic affective priming effect could not be demonstrated in a few studie s. But, the complete data suggest that the effect under investigation is rather pervasive and unconditional. In addition, it is important to note that the evidence for automatic stimulus evaluation, derived from the affective priming paradigm, is not restricted to speci® c classes of stimuli, but can be generalis ed towards different types of stimulus materials, like words (Bargh et al., 1992 ;
Chaiken & Bargh, 1993; Fazio et al., 1986; Hermans et al., 1994), simple
line drawings (Giner-Sorolla, Garcia, & Bargh, 1994), and complex real
life colour pictures (Hermans et al., 1994, 1996). It has also been obtained
for a wide range of stimuli varying in content (Hermans, 1996 ), as well as in accessibili ty of their affec tive valence in memory (Bargh et al., 1992, 1996 ; for further discussion, see Chaiken & Bargh, 1993; De Houwer, Hermans, & Eelen, 1998; Fazio, 1993). However, although all models assume that automatic stimulus evaluation (and hence affective priming) should apply to all sensory modalitie s, all studie s publis hed on the affec tive priming effect have in common that the stimulus material is of visual nature. In this context, it was decided to use odours as primes in an affective priming procedure. Accordingly , to test whether the affective priming effect can be generalised to nonvisual stimuli, a cross-modal affective priming paradigm was implemented, in which individuall y selected positive and negativ e odours were used as an affective processing context for word evaluation.
METHOD
Subjects
Forty-thre e students (27 males, 16 females) from various departments
(mainly Psychology and Economic s) of the University of Leuven volunteered
to partic ipate in the experime nt.
AFFECTIVE ODOUR PRIMING 603
Stimuli
Prime s were odours which were selected on an a-priori basis for their
affective qualitie s. The ® nal set of ten odours consisted of seven singular
essential oils (lemon, hops, peppermint, bitte r almond, Dauc us carota,
Sassafras albidum, and hyssop), two essential oil complexes (lavender,
rosemary, and thyme; roses, pine , cinnamon, incense, and mandarin), and pure civet. The latter was dissolved in ethyl alcohol (94%) to obtain good evaporation. Targets were Dutch nouns, which were selected from the Hermans and De Houwer (1994) affective norms for 720 Dutch words. Based on their mean affective rating in this normative study, 20 positive , 20 negative , and 10 neutral words were selected for the target selection procedure (see Appendix) .
Apparatus
The odours were presented by means of an olfactometer which was
constructed for the purpose of this experiment. It consisted of a wooden
box in which three plas tic holders were attached, each containing a glass
jar with plastic screw-on lid. A PVC tube conne cted a bottle of compressed
oxygen to each jar, from which a second PVC tube lead to one of four
electromagnetic valve s (Danfoss , Type EVR 3, 46 bar), which were controlled
by a 486 DX± IBM compatible computer. During the priming
procedure, one jar contained a positive odour, and the second jar a
negative odour, which were both selected by the subje ct. The third jar
contained a mixture of one part of water and one part of vinegar. This
solution was used to clean the plas tic tubing , and to neutralise the nose for
the next odour presentation.
During inter-trial intervals (ITI) the main valv e was open, whereas the
three other valve s, connected to the jars with the odours or the vine gar
solution, remained closed. During odour presentation, the main valv e
closed, immediately followed by the opening of one of the other valve s,
thereby letting the oxygen circulate through one of the jars containing an
odour. The odour-saturate d air was then transported from the olfactometer
to the subje ct.
During the actual odour priming procedure, each subje ct wore a safety
headphone , which preve nted hearing the opening and closing of the
electromagnetic valve s. Both the plas tic tube that transported the saturated
air from the olfactometer, and the voic e key microphone were attached to
the headphone . Their position was adjustable . Reaction time data were
recorded by means of a voice key that was connec ted to the computer
which also controlled the presentation of the odours and the target words.
604 HERMANS, BAEYENS, EELEN
Procedure
First, the subje cts were asked to read the instruction sheet. It stated that we
were inve stigating the general impact of environmental factors on the
cognitiv e processing of words. It was explaine d that this study was part of a large r series of experiments, in which other factors, such as the effects of music and concurrent mental loads, had been explore d. The present study was introduc ed as a further inve stigation of the general in¯ uence
of odours on the perception and processing of words. The experiment took place in two different rooms. In the ® rst room, target words and odourprimes were selected on an individual basis. The actual priming phase , and an odour discrimination task were conducted in a second room. During the ® rst phase of the experiment, target words were selected by each individual subje ct (target selection phase). The experimenter handed
over a set of 50 cards on which a positive , negative , or neutral noun was printed. After having gone quic kly through this set to get an overall impression of the type of words included, subjects evaluated each of the 50 words separately, by laying them down on a 21-category scale, ranging from 2 100
(very negative ), over 0 (neutral ), to +100 (very positiv e). It was stressed to
rate the words, relying on an immediate and spontaneous impression. Next (prime selection phase ), the subje ct was led to anothe r table on
which there were 10 small plas tic containers, each containing a pie ce of cottonwool on which one of the 10 odours was pipetted. The subje ct was asked to open each containe r, to smell it, to close it again, and to evaluate the odour by placing the containe r on the left side of the table (negative odours), in the middle (neutral odours), or on the right side of the table (positiv e odours). Following this crude categorisation, the subje ct was asked to smell the odours for a second time, and to select the odour he/ she liked most, and the odour they liked least. Meanwhile, based on the subje ct’ s affective ratings of the nouns, the experimenter selected the 10 most positive and the 10 most negativ e words. These were typed into the computer program, and then the selected positiv e and negative odour were attached into the olfactometer. Next, the subject was lead to the second room for the actual priming procedure (odour priming phase ), and was asked to sit in front of the
computer screen at a distanc e of about 50cm. It was explaine d that with an interval of about half a minute , words would appear on the computer screen which they had to evaluate as quickly as possible , by saying out loud either ``pos itive’ ’ or ``negative’ ’ . The function of the voic e key was
explaine d in detail. Next, the headphone with microphone and odour tube was adjusted to the subje ct’ s head. The open ends of the plastic tube were positione d about 1cm under the nostrils of the partic ipant. The oxyg en was decompressed to a continuous ¯ ow of about 13 litre s a minute.
AFFECTIVE ODOUR PRIMING 605
In total, the priming phase consisted of two blocks of 20 trials (Block 1/
Block 2). Within each block, each of the 20 target words was presented
once . Presentation order was randomised for each block and for each
subject separately. Within each block, both positive and negative odour
primes were presented 10 times, and were allocated randomly to the target
words, the only restriction being that there should be ® ve presentations for
each of the four prime-target combinations . Accordingly, each block
consisted of 10 affectively congruent trials (® ve positiv e-positive , and
® ve negative -negative prime -target trials), and 10 affectively incongru ent
trials (® ve negative -positive , and ® ve positive -negativ e trials).
Each trial started with the presentation of the odour during 10sec, immediate ly followed by a row of six ® xation crosses in the centre of the screen, and a short warning tone (200msec, 1000Hz ) to attract the subject’ s attention to the computer screen. Tone and crosses were immediate ly replac ed by the target word, which was presented in white uppe rcase letters (height 8mm) against an all-black background in the centre of the screen. The target word remained on the screen until the response was registered by the voic e key; with a maximum presentation time of 2000msec. The inter-trial interval (ITI) was always 23sec. During the first six seconds of the ITI, oxyg en saturated with the vinegar solution was circulated through the plastic tubes to remove traces of the previous odour. The two blocks of 20 experimental trials, which were separated by a short break, were preceded by four practice trials to familiari se the subje ct with the procedure [ target words: party (feest), sadis t (sadist), blossom (bloe -
sem), and funeral (begrafenis)] . The ® rst two practice trials included the
positive odour as prime. In the third and fourth practice trial the negativ e odour prec eded the targe t word. The experiment concluded with an odour discrimination task. This was
implemented as a test of the subje ct’ s ability to discriminate between the two experimental odours used in the odour priming phase . Three plastic containe rs were placed in front of the subje ct, two of which contained the same odour (one of the two odour primes). The third containe r held the other experimental odour. The subje ct was asked to open each containe r, to smell it and select the deviant odour. This test was repeated with the other expe rimental odour as the deviant stimulus. Subjec ts who failed to dis - criminate between the odours (one of two misses), were excluded from the analyse s.
RESULTS
The data of two subje cts were discarded from the analyse s because of
technic al problems. The data of a third subje ct were discarded because
she failed at one of both trials of the odour discrimination task. For the
606 HERMANS, BAEYENS, EELEN
remaining 40 subje cts (25 males, 15 females), all response latencies large r
than 1500ms ec, or shorter than 250msec were considered as outlie rs
(1.13%). Based on all remaining data, means were calculated for affectively
congrue nt trials (positive -positiv e; negativ e-negative ), and affectively
incongruent trials (positiv e-negative ; negative -positive ). This was
done separate ly for the ® rst (trials 1± 20), and the second (trials 21± 40)
experimental block.
These data were analysed in a 2 (Gender: Male/Female) 3 2 (Block:
Block 1/Block 2) 3 2 (Target: Positive /Negative ) 3 2 (Affective Congruence:
Congrue nt/Incongruent) ANOVA with repeated measurements for the last three variable s1. As predicted, this analysis showed a signi® cant
main effect of Affective Congruence [F(1, 38) = 4.89 ; P < .05] . Target
words were evaluated signi® cantly faster when preceded by a similarly valenced odour prime (M = 606msec), as compared to trials for which
prime and target were of opposite valence (M = 616msec).
In addition to this effect of Affective Congrue nce, the analys is showed a signi® cant main effect of Block [F(1, 38) = 10.64; P < .005; Mblock 1 = 626;
Mblock 2 = 596] , which is a standard observation in most affective priming
studie s (Hermans, 1996), and can be considered as an effect of practice. There was also a main effect of target valence [F(1, 38) = 18.72; P <
.0005] . Positive targets were evaluated faster (M = 591) than negative
targets (M = 631). Response latencies did not differ between males and
females [F(1, 38) < 1; n.s.; Mmales = 603; Mfemales = 625] .
As shown in Fig. 1, the effect of Affective Congruence was moderated by Gender [F(1, 38) = 6.97; P < .02] . For female subje cts (N = 15), the
means were clearly in line with the affective priming hypothe sis [Mcongruent
= 609; Mincongruent = 641] . For the male group (N = 25), however, there was
no difference between response latencies for affectively congrue nt and incongruent trials [Mcongruent = 604; Mincongrue nt = 601] . This was con-
® rmed by separate 2 (Block) 3 2 (Target) 3 2 (Affective Congruence)
analyse s for each of both groups. For the female subgroup, the main effect of Affective Congruence was signi® cant [F(1, 14) = 7.87; P < .02] . But this
effect was absent for the male subje cts [F(1, 24) < 1; n.s.] . However, there
was a marginally signi® cant Block 3 Affective Congruence interaction for
the latte r group [F(1, 24) = 3.24; P = .085 ] . For the ® rst 20 trials, response
AFFECTIVE ODOUR PRIMING 607
1 In a preliminary analysis, in which Gender was not included as a between-subjects
variable , the main effec t of Affec tive Congruence was only marginally signi® cant [F(1, 39)
= 2.22; P = .14; MCong rue nt = 606; MIncong rue nt = 616] . But, because there was a signi® cant
correlation between a general measure of affective priming and Gender [ r(40) = .39; P < .02] ,
it was decided to include Gender as a between-subjects variable in the analysis. Because both ANOVAs produced comparable results, it was decided to report only the one inc luding Gender.
latenc ies were faste r for affectively incongrue nt trials (M = 611) as
compared to affectively congrue nt trials (M = 628). In the second half of
the experiment, however, means were in line with the affective priming hypothesis (Mcongruent = 580; Mincongruent = 592). A similar interaction was
absent for the female subje cts (F < 1).
Finally, because a signi® cant main effect of the valence of the prime stimulus was observed in several previous affective priming studies (Hermans, 1996 ), an additional analysis of variance, with prime valence as a third within-subje cts variable , was conducted [2 (Gender) 3 2 (Block) 3 2
(Prime: Positive /Negative ) 3 (Affective Congrue nce) ANOVA] . Indeed,
this analy sis revealed a signi® cant main effect of Prime [F(1, 38) = 9.59;
P < .005] . After a negativ e odour (M = 619), subje cts responded about
15msec slower than after a positive odour (M = 604).
DISCUSSION
In a recent series of affective priming studie s (Bargh et al., 1992 , 1996 ; De
Houwer et al., 1998 ; Fazio et al., 1986; Greenwald et al., 1989, 1995;
Hermans, 1996; Hermans et al., 1994 , 1996) it has been demonstrated that
the affective valence of a prime stimulus in¯ uences the speed of response s
towards affectively congruent or incongrue nt target stimuli. These data
have been interpreted as evidence for the assumption that humans are
endowed with an evaluative decision mechanism which allows them to
automatically evaluate afferent stimulus information (Zajonc, 1980 ). The
608 HERMANS, BAEYENS, EELEN
FIG. 1. Mean evaluative response latencies for affectively congruent and incongrue nt trials as a
function of Gender (Female/Male).
generality of this affective priming effect is now well established, and has
been validated using different types of stimuli and procedural variations.
Nevertheless, one aspect shared by all studie s published on affective priming
is the visual nature of the priming stimuli. In this sense, the present
experiment can be regarded as a further generalisation of the basic observation,
as the data show that olfactory primes speeded up evaluative response
latencies for affectively congrue nt targets as compared to affectively incongruent
trials , thus replic ating the results of prior affective priming studie s
with visual priming stimuli. Not only are odours of a nonvis ual nature; in
comparison with the types of stimulus materials used in previous studie s,
they are also an outstanding example of external stimuli for which there is
only a very limited role of linguisti c proc esses (Richardson & Zucco, 1989).
As a consequence, the ability to recognise and identify an individual odour
is extremely limited (Schab, 1991 ). On the other hand, odours can have a
considerable affective impact (Ehrlichman & Halpern, 1988; Kirk-Smith,
Van Tolle r, & Dodd, 1983 ; Miltne r et al., 1994). Furthermore, it has to be
noted that this study is the ® rst demonstration of cross-modal affective
priming (olfactory primes, visual targets). Until now, all other publishe d
studie s have used primes and targets of the same sensory modality .
The observed affec tive priming effect was, however, only pre sent for the
female subgroup. For the male subje cts, there was no difference between
affectively congruent and affectively incongrue nt trials. Although it is the
® rst time that gender effects are reported in the context of affective
priming, it does not completely come as a surprise (e.g. Kirk-Smith et
al., 1983), as it might simply be attributed to the fact that the female
olfactory sense is superior in most respects to that of males (Doty, 1991 ;
Doty, Applebaum, Zusho, & Settle, 1985). Thresholds for odour identific ation
are much lower in women for several odours ; odours are experienced
as more intense by females; they have a better capacity to discriminate
odours; and females are more pronounc ed in their judgement of whether a
partic ular odour is pleasant or unple asant (Vroon, Van Ameronge n, De
Vries, 1994, p. 95). This would imply that the absence of affective priming
for males is not a consequence of their inability to evaluate automatically
odours, but of their impaired capacity to detect and/or discriminate odours.
The fact that the means for the males were in the expected direction in the
second experimental block (see the near-signi® cant Block 3 Affective
Congrue nce interaction) supports this hypothes is. The olfactory system of the male subje cts probably needed more time to perceive the odour primes. Indeed, the conce ntration of the odours in the presented air was very unobtrusive . And although a vinegar solution was used to clear the plastic tubes after each trial, it was impossible to obtain a complete deodorisation. As a result, remnants of previous trials might have lead to a diminishe d perception of the following odours. Therefore, the use of
AFFECTIVE ODOUR PRIMING 609
more trials or less unobtrusive odours would probably lead to similar
results for males.
As well as the main effect of affective congruence, two other ® ndings
which are common in affective priming research could be replicated in the
present study, namely the main effects of Target Valence and Prime
Vale nce. Response latencies towards negative target words were slower
as compared to positive target words, regardle ss of the valenc e of the prime
stimulus , and slower response latencies were obse rved after negative than
after positive odour primes, regardle ss of target valence. In several of our
previous studie s, similar effects have been obtaine d. These main effects
should, however, not nec essarily be attributed to differences in affective
value , but can also be ascribed to difference s in affective extremity or
subjective familiari ty. Nevertheless, we have observed similar main
effects, even if we controlled for such variable s (Hermans, 1996). This
would suggest that the faster response latencie s after positive prime s or to
positive targets are at least in part of affective nature. But, whatever their
nature, it is worthwhile reporting that a posteriori analyse s revealed that
whereas the main effect of Target Valence was present for both female
[F(1, 14) = 9.14; P < .01] and male subje cts [F(1, 24) = 10.64 ; P < .01] , the
main effect of Prime Valence was restricted to the female partic ipants [F(1,
14) = 8.45; P < .02; Males: F(1, 24) = 1.85; P = .18; Gender 3 Prime
Vale nce: F(1, 38) = 1.98 ; P = .17] . This is a further indic ation for the
hypothesis that our male subje cts were simply less able to perceive the rather unobtrusive experimetal odours. Regarding this issue, it is worth noting that the perception of an odour, which is based on a contact between the odour molecules and the relevant sensory structures, at least require s some minimal inhalation of these odour molecules through the nose. To ensure that the subjects would have the chance to inhale and perceive the odours , the primes were presented for a duration of 10 seconds. As the target immediately followed after the prime (ISI = 0), the SOA in this study was 10sec, which is a marked difference from the SOA of 300msec (Prime = 200msec; ISI = 100msec), which is most often used in affective priming research. At ® rst glanc e, the signifi - cant priming effect in this study may be rather unexpected, given the earlier mentioned studie s that demonstrated that af fective priming effects disappear when longer SOAs (450msec, 1000msec) are used. The crucial variable in these SOA studie s, however, is not the SOA (prime duration plus ISI), but the duration of the ISI (with prime duration held constant). It is assumed that the automatic af fective in¯ uence of the prime is only shortlived because it is worn out or actively inhibite d at the moment of target presentation when long ISIs are used. In this sense, the procedure of the present experiment is more comparable to other studie s that obtained
610 HERMANS, BAEYENS, EELEN
signi® cant affective priming effects using zero or very short ISIs (Hermans,
1996 ).
Hence, the present priming effect is interpreted as the product of a
passive , automatic activation process, rather than as the result of more
strategic, conscious processes. This interpretation is supported by the fact
that strategic processes, in the sense of expectancy generation (Neely,
1991 ; Shelton & Martin, 1992), are unlike ly to occur in the present
priming procedure because the valence of the prime s was in no way
predictive of the valence of the target. Moreover, the employme nt of
such conscious, strategic processes would also be counterproductive
because these processes are assumed to be time-consuming, and because
the primary task (target evaluation) in our experiment was only a very
simple one.
Manuscript received 9 April 1997
Revised manuscript received 10 July 1997
REFERENCES
Bargh, J.A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the attitude
activation effe ct. Journal of Personality and Social Psychology, 62, 893± 912.
Bargh, J.A., Chaiken, S., Raymond, P., & Hymes, C. (1996). The automatic evaluation
effect: Unconditional automatic attitude activation with a pronunciation task. Journal of
Experimental Socia l Psychology, 32, 104± 128.
Chaiken, S., & Bargh, J.A. (1993). Occurrence versus modi® cation of the automatic activ ation effect: Reply to Fazio. Journal of Personality and Social Psychology, 64, 759± 765.
De Houwer, J., Hermans, D., & Eelen, P. (1998). Affectiv e and identity priming with episodically associated stimuli. Cognition and Emotion, 12, 145± 169.
Doty, R.L. (1991). Psychophysical measurement of odor perception in humans. In D.G.
Laing, R.L. Doty, & W. Breipohl (Eds.), The human sense of smell. Berlin: Springer.
Doty, R.L., Applebaum, S., Zusho, H., & Settle , R.G. (1985). Sex differences in odor
identi® cation ability. A cross-cultural analysis. Neuropsychologia, 23, 667± 672.
Ehrlichman, H., & Halpern, J.N. (1988). Affe ct and memory: Effects of pleasant and
unpleasant odors on retrieval of happy and unhappy memories. Journal of Personality
and Social Psychology, 55, 769± 779.
Fazio, R.H. (1993). Variability in the like lihood of atuomatic attitude activation: Data
reanalysis and commentary on Bargh, Chaiken, Govender, and Pratto (1992). Journal
of Personality and Socia l Psychology, 64, 753± 758.
Fazio, R.H., Sanbonmats u, D.M., Powell, M.C., & Kardes, F.R. (1986). On the automatic
activation of attitudes. Journal of Personality and Social Psychology, 50, 229± 238.
Giner-Sorolla, R., Garcia, M.T., & Bargh, J.A. (1994). The automatic evaluation of pic -
tures. Unpublished manuscript, New York University, New York.
Greenwald, A.G., Klinge r, M.R., & Liu, T.J. (1989). Unconscious processing of dichoptically masked words. Memory and Cognitio n, 7, 35± 47.
Greenwald, A.G., Klinger, M.R., & Schuh, E.S. (1995) . Activation by marginally perceptible (``subliminal’ ’ ) stimuli: Dissociation of unconscious from conscious cognition.
Journal of Experimental Psychology: General, 124, 22± 42.
Hermans, D. (1996). Automatisc he stimulusevaluatie. Een experimentele analyse van de
AFFECTIVE ODOUR PRIMING 611
voorwaarden voor evaluatiev e stimulusdiscriminatie aan de hand van het affectieveprimingparadigma.
[Automatic stimulus evaluation. An experimental analysis of the preconditions for evaluative stimulus discrimination using an affective priming paradigm] . Unpublished doctoral dissertation, University of Leuven, Belgium. Hermans, D., & De Houwer, J. (1994). Affec tive and subje ctive familiarity ratings of 740 Dutch words. Psychologica Belgica, 34, 115± 139.
Hermans, D., De Houwer, J., & Eelen, P. (1994). The affective priming effect: automatic activation of evaluative information in memory. Cognition and Emotion, 8, 515± 533.
Hermans, D., De Houwer, J., & Eelen, P. (1996). Evaluative decision latencies mediated by
induced affective states. Behaviour Re search and Therapy, 34, 483± 488.
Hill, A.B., & Kemp-Whe eler, S.M. (1989). The in¯ uence of context on lexical decision
times for emotionally aversive words. Current Psychology: Re search and Re views, 8,
219± 227. Kirk-Smith, M.D., Van Toller, C., & Dodd, G.H. (1983). Unconscious odour conditioning in humans. Biological Psychology, 17, 221± 231.
Klauer, K.C., Rossnagel, C., & Musch, J. (1997). List-contex t effects in evaluativ e priming.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 23, 246± 255.
Miltner, W., Matjak, M., Braun, C., Diekmann, H., & Brody, S. (1994). Emotional qualitie s
of odors and their in¯ uence on the startle re¯ ex in humans. Psychophysiology, 31, 107±
110.
Neely, J.H. (1977). Semantic priming and retrieval from lexical memory: roles of inhibitionle
ss spreading activation and limited-capacity atte ntion. Journal of Experimental
Psychology: General, 106, 226± 254.
Neely, J.H. (1991) . Semantic priming effects in visual word recognition: A selective review of current ® ndings and theories. In D. Besner & G.W. Humphreys (Eds.), Basic processes
in reading: Visual word recognition (pp. 264± 336). Hillsdale, NJ: Lawrence Erlbaum
Associates Inc. Richardson, J.T.E., & Zucco, G.M. (1989). Cognition and olfaction. Psychological Bulletin,
105, 352± 360.
Schab, F.R. (1991). Odor memory: Taking stock. Psychological Bulletin, 109, 242± 251.
Shelton, J.R., & Martin, R.C. (1992). How semantic is automatic semantic priming?
Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 1191± 1210.
Vrana, S.R., Spence, E.L., & Lang, P.J. (1988). The startle probe response: a new measure
of emotion? Journal of Abnormal Psychology, 97, 487± 491.
Vroon, P., Van Amerongen, A., & De Vries, H. (1994). Verborgen verleider. Psychologie
van de reuk. [Hidden seducer. Psychology of smell] . Baarn: AMBO.
Wentura, D. (1996). The ``meddlin g-in’’ of affec tive information: Evidence for negative
priming and implic it judgement tendencies in the affe ctive priming paradigm. Unpublished
manuscript, University of MuÈ nster, Germany. Zajonc, R.B. (1980). Feeling and thinking: Preferences need no infe rences. American
Psychologist, 35, 151± 175.
612 HERMANS, BAEYENS, EELEN
AFFECTIVE ODOUR PRIMING 613
APPENDIX
DUTCH WORDS USED IN THE TARGET SELECTION
PHASE
Positive Negative Neutral
liefde (love) moord (murder) tand (tooth)
lach (laugh) verkrachting (rape) boter (butter)
kus (kiss) incest (inc est) kever (beetle )
vriend (friend) oorlog (war) ivoor (ivory)
vakantie (vacation) aids (aids) sprinkhaan (grasshopper)
vrede (peace) marteling (torture) vrachtwagen (truck)
omhelzing (embrace) tumor (tumour) slaapzaal (dormitory)
trouw (marriage) executie (execution) vierkant (square)
zomer (summer) ziekte (illness ) strijkplank (ironing-board)
knuffel (hug) kanker (cancer) schaar (scissors)
zon (sun) haat (hatre d)
romantiek (romance) Hitler (Hitle r)
muziek (music) alcoholisme (alcoholism)
thuis (home) gezwel (growth)
humor (humour) verstikking (suffocation)
bloemen (¯ owers) misdaad (crime)
lente (spring) geweren (guns)
geschenk (gift) slachting (slaughter)
verrassing (surprise) ongeluk (accide nt)
warmte (warmth) coma (coma)