The Emotion Probe

Studies of Motivation and Attention

Peter J. Lang
University of Florida

Emotions are action dispositions—states of vigilant aversive system's brain circuitry in conditioned fear and
readiness that vary widely in reported affect, physiology, specifies its autonomic and somatic output paths. Of par-
and behavior. They are driven, however, by only 2 opponent ticular interest is the manner in which an active aversive
motivational systems, appetitive and aversive—subcortical system modulates concurrent but independently evoked
circuits that mediate reactions to primary reinforcers. Us- behaviors. The neurophysiology of this phenomenon is
ing a large emotional picture library, reliable affective well understood for the startle reflex, which, in animals,
psychophysiologies are shown, defined by the judged va- is strongly potentiated (is larger) when aversive (fear) mo-
lence (appetitive/pleasant or aversive/unpleasant) and tivational circuits are activated.
arousal of picture percepts. Picture-evoked affects also The subsequent exposition is centered on research us-
modulate responses to independently presented startle ing the startle reflex to probe the affective states of human
probe stimuli. In other words, they potentiate startle re- beings. These experiments test a theory of motivational
flexes during unpleasant pictures and inhibit them during priming or, more specifically, two main corollary hy-
pleasant pictures, and both effects are augmented by high potheses: (a) Defensive reflexes, including startle, increase
picture arousal. Implications are elucidated for research in amplitude when an organism is aversively motivated
in basic emotions, psychopathology, and theories of ori- (i.e., when the aversive system is activated and the indi-
enting and defense. Conclusions highlight both the ap- vidual's emotional state is affectively unpleasant); and (b)
proach's constraints and promising paths for future study. Defensive reflexes are reduced in amplitude when an or-
ganism is positively motivated (i.e., when the appetitive
system is activated and the individual's emotional state
is affectively pleasant).

T he purpose of this discourse is to describe a scientific

view of emotion that can be explored productively
in the laboratory and yet provide a framework for
studying practical concerns about emotion in society, the
media, and in affective disorder. It begins with a working
The data show very wide support for these hypotheses:
Using acoustic probe stimuli and measuring the early
blink component of the startle response, the reflex aug-
mentation found in animal studies with shock sensiti-
zation and conditioning (anticipation) has been replicated
definition of emotions as action dispositions, describing
in humans. Furthermore, when humans look at pleasant
a database that includes functional behaviors, evaluative
and expressive language, and physiological events. These
responses all vary enormously within and between emo- Editor's note. Articles based on APA award addresses that appear in
tions; however, it is put forward that this response com- the American Psychologist are scholarly articles by distinguished con-
plexity is orchestrated by simpler, underlying motivational tributors to the field. As such, they are given special consideration in
the American Psychologist's editorial selection process.
parameters. All emotions can be located in a two-dimen- This article was originally presented as part of a Distinguished
sional space, as coordinates of affective valence and Scientific Contributions award address at the 102nd Annual Convention
arousal. This dimensional conception is applied here to of the American Psychological Association in Los Angeles, California.
a perceptual task, organizing an analysis of human affec- Melissa G. Warren served as action editor for this article.
tive reactions to emotionally evocative pictures.
Author's note. The theory and findings presented here are largely a
The concepts of affective valence and arousal are ex- distillation of effort by a team composed of Margaret M. Bradley, Bruce
plicated in terms of specific motivational systems in the N. Cuthbert, and Peter J. Lang, all working at the National Institute of
brain. Following Konorski (1967) and others (e.g., Dick- Mental Health Center for the Study of Emotion and Attention (NIMH-
CSEA) at the University of Florida. This work was supported in part by
inson & Dearing, 1979; Solomon & Corbit, 1974), it is National Institute of Mental Health Grants ROI MH37757, R01
suggested that affects are driven by two primary motive MH41950, and R01 MH43975, Behavioral Sciences Research Center
systems: the appetitive system (consummatory, sexual, Grant P50 MH52384, and Grant R01 AGo9779 from the National
and nurturant), prototypically expressed by behavioral Institute of Aging.
approach, and the aversive system (protective, withdraw- CSEA investigators Michael Davis and John Cacioppo provided im-
portant consultation. Thanks are also due to the many past colleagues
ing, and defensive), prototypically expressed by behavioral and student collaborators whose work helped shape these views.
escape and avoidance. With respect to the latter, recent Correspondence concerning this article should be addressed to Peter
animal research is reviewed that (using the tools of phar- J. Lang, Box 100165 H.S.C., Department of Clinical and Health Psy-
macological and neurosurgical mapping) defines the chology, Gainesville, FL 32610-0165.

372 May 1995 • American Psychologist

Copyright 1995 by the American Psychological Association, Inc. 0003-066X/95/S2.00
Vol. 50, No. 5, 372-385
emotional scenes, concurrent startle reflexes are inhibited; we live. Elaborate instrumental acts, delay, and inhibition
the same evoked reflexes are augmented when they occur have evolved to facilitate the goal-directed behavior of
in the context of unpleasant picture stimuli. These recip- more complex organisms. Thus, a person's escape from
rocal effects for pleasant and unpleasant stimuli become aversive stimuli may be achieved as well by attack or
stronger as both types of stimuli are judged to be more compliance as by flight, and circumspection often rewards
arousing. an appetite that a direct approach could not satisfy.
Motivational modulation of the startle reflex is gen- It is clear that the contextual tactics of approach and
erally robust despite habituation, variations in probe avoidance have become more varied in humans; nev-
timing and in specific foreground content, or changes in ertheless, the strategic frame of appetite and aversion is
probe intensity and modality. Furthermore, it appears to no less relevant. Emotions are products of a Darwinian
be specific to startle in the perceptual context, and similar development and could be characterized as motivationally
modulation is not observed either for other tested reflexes tuned states of readiness. In humans, the presumed in-
or for the cortical event-related potential (P300) evoked dices of these "affects" include responses in three reactive
by the same probe stimulus. systems: (a) expressive and evaluative language, (b) phys-
The replicability of startle probe effects commends the iologic changes mediated by the somatic and autonomic
method for use in studying special subject populations as systems, and (c) behavioral sequelae, such as patterns of
a way of assessing affective state: Research suggests that avoidance or performance deficits. This is the database
emotional reflex modulation appears very early in human of emotion, and a theory of emotion must cope with its
development and persists with aging. There is evidence breadth and diversity. The task is complicated by the fact
that when phobic individuals confront phobic objects, that the correlations among and within systems are often
they show greater than normal reflex potentiation. On quite modest, for example (Lang, 1968; Mandler, Mand-
the other hand, when facing stimuli they describe as aver- ler, Kremen, & Sholiton, 1961), when research partici-
sive, psychopathic prisoners fail to show normal startle pants confront a situation designed to evoke a particular
potentiation. Furthermore, studies of anxiety disorders affect (e.g., anger, fear, and disgust). Furthermore, the
show significant differences in startle potentiation ac- patterns of response that are observed are often unreliable
cording to diagnosis (e.g., phobic patients and patients within subjects and across different contexts of stimula-
with post-traumatic stress disorder show greater poten- tion (Lacey, 1958; Lacey & Lacey, 1970). Thus, emotional
tiation than patients with panic disorder). These data raise judgments, physiology, and behavior can present a con-
questions about the varying strength of inhibitory and fusing rock pile that resists a simple classification by spe-
activating circuits in the brain as they modulate attention cific emotional states.
and emotional expression.
Affective Valence and Arousal
Emotions Are Action Dispositions Although emotional expression is highly varied, many
William James (1894) once said that emotions are as var- theorists view its motivational basis as having a much
ied as the shapes of rocks on a New Hampshire farm. simpler, two-factor organization. Konorski (1967, p. 9)
Unlike New Hampshire rocks, however, they are not inert advocated a biphasic model that he founded on a typology
granite but more like a protean lava. Emotions are about of unconditioned reflexes. Exteroceptive reflexes were
doing something that is important to the organism (es- seen to fall into two classes, preservative (e.g., ingestion,
cape, attack, sexual consummation, etc.). On the other copulation, and nurture of progeny) and protective (e.g.,
hand, curiously, the defining acts may or may not actually withdrawal from or rejection of noxious agents) on the
occur. basis of their biological, motivational role. Differentiating
Emotions are systemic responses that happen when his views from Hess (1957), Konorski stressed that acti-
highly motivated actions are delayed or inhibited. Or, as vation or arousal modulated both preservative and pro-
Hebb (1949) proposed, emotions result when novel cir- tective reactions. These reactions were considered to be
cumstances prevent completion of cued behavior. Thus, the behavioral foundation of affects and expressed emo-
emotions quintessentially occur in a behavioral hiatus, tions. Dickinson and Dearing (1979) developed Konor-
as states "experienced," then reported on and evaluated ski's dichotomy into two opponent motivational systems,
(see Frijda, 1986, on "feelings"). It is in this sense that aversive and attractive, each activated by a different but
they are dispositions and not the acts themselves: They equally wide range of unconditioned stimuli. These sys-
reflect central activation and preparation for action. tems were held to have "reciprocal inhibitory connec-
It is likely that the varied shapes of emotions alluded tions" (p. 5) that modulate learned behavior and responses
to by James (1894) have evolved from simpler action ten- to new, unconditioned input.
dencies. The behavior of very primitive organisms can be Since Wundt (1896), the view that affects might be
wholly characterized by two responses—a direct approach organized by overarching motivational factors has also
to appetitive stimuli and withdrawal from nociceptive been suggested by researchers studying the emotion re-
stimuli (see Schneirla, 1959). This modest behavioral ports of humans. Thus, work on natural language cate-
repertoire, sufficient to a flatworm, cannot of course im- gories (Ortony, Clore, & Collins, 1988; Shaver, Schwartz,
plement the many subgoals of humans nor manage the Kirson, & O'Connor, 1987) suggests that people's knowl-
perceptually richer, more complex environment in which edge about emotions is hierarchically organized and that

May 1995 • American Psychologist 373

the superordinate division is between positivity (pleasant can be rapidly administered (see Figure 1). SAM assess-
states: love and joy) and negativity (unpleasant states: an- ments of pleasure and arousal generally correlate .9 and
ger, sadness, and fear). Osgood and his associates (e.g., above with the measures of these judgments, using Mehr-
Osgood, Suci, & Tannenbaum, 1957), using the semantic abian and Russell's (1974) semantic differential scales (see
differential, showed that emotional descriptors could be Bradley & Lang, 1994).
in the main distribution along a bipolar dimension of A representative sample of IAPS pictures, distributed
affective valence—from attraction and pleasure to aver- in valence/arousal space, is presented in Figure 1. The
sion and displeasure. A dimension of activation—from locations of typical picture content are shown according
calm to aroused—also accounted for very significant to coordinates from the initial standardization study. To
variance. Similar conclusions have been drawn by other help orient the reader, the SAM locations of various emo-
investigators of verbal report (e.g., Mehrabian & Russell, tional words noted are obtained from independent rating
1974; Russell, 1980; Tellegen, 1985) and in early studies experiments. The overall boomerang shape of the picture
of facial expression (Schlosberg, 1952). mass (with its two arms seeming to extend from a com-
The present view integrates these several lines of theory mon calm, no-affect base, towards either high arousal
development. It is proposed that two motive systems exist pleasant or high arousal unpleasant locations) is not in-
in the brain—appetitive and aversive—accounting for the consistent with an underlying bi-motivational organiza-
primacy of the valence dimension. Arousal is not viewed tion. Despite efforts to fill its peripheral gaps, this form
as having a separate substrate, but rather, as reflecting of the emotion-perceptual space has remained stable over
variations in the activation (metabolic and neural) of ei- several years of picture collection and research.
ther or both systems (see also Cacioppo & Berntson, In studies of these picture stimuli, SAM-based self-
1994). Tactical demands of context may variously shape evaluative judgments have shown good stability (Green-
affects. All affects are, however, organized around a mo-
tivational base. In this sense, valence and system arousal
are the strategic dimensions of the emotion world.
Looking at Pictures Figure 1
Distribution of 360 Photographic Images From the
The emotional stimuli that are the focus of this presen- International Affective Picture System (IAPS)
tation are pictures. An effort has been made to collect
photographs of people, animals, nature, objects, events, International Affective Picture System
and scenes, sampling as broadly as possible the range of
visual representations of the world. The reactions evoked
by these picture stimuli do not, of course, mirror all the
affects that occur in life. On the other hand, media rep-
resentations are a central feature of current culture, and
icons of religion, family values, beauty, natural disasters,
pornography, crime, and violence—in art, film, maga-
zines, and television—are widely held to be powerful
emotion generators and significant determinants of so-
cietal manners and mores.
Pictures are, furthermore, convenient laboratory stim-
uli that permit controlled exposure, in timing and inten-
sity, and exact reproduction within and between experi-
ments and laboratories. An aim of this effort is to produce
calibrated emotional stimuli that can serve as a mea-
surement standard—analogous to those used in physical
metrics—and thus encourage scientific replication in re-
search on emotion.
In the laboratory studies to be described, emotional
stimuli were presented either as photographic slides or,
when digitized, as displays on a computer monitor. Cur-
rently there are over 400 pictures in the International Arousal

Affective Picture System (IAPS; Center for the Study of Note. Images are organized in a two-dimensional space, defined by the judged
dimensions of valence and arousal. Specific picture contents are indicated for the
Emotion and Attention, 1994). Pictures are initially stan- images denoted by filled circles. Standardization samples of approximately 100
dardized in groups of 60 slides, using participant samples individuals used the Self-Assessment Manikin ISAM; Lang, 1980) to make these
of approximately 100. In this initial calibration process, judgments. In its paper-and-pencil version, SAM is a 9-point scale, with valence
and arousal represented graphically by changes in a cartoon figure I© Peter
participants rate their emotional experience of each pic- Long; see x,y axes above). SAM correlates highly with the semantic differential
ture on scales of affective valence and arousal, using the estimate of these same dimensions (Bradley & Lang, 19941. SAM ratings of various
emotion words are located in the same space (see words in italics, in boxes).
Self-Assessment Manikin (SAM; Lang, 1980). This in-
strument is pictographic in form, largely culture free, and

374 May 1995 • American Psychologist

 Thus, for example, when subjects discriminate newly
Figure 2 presented from previously seen pictures, they key press
Covariation of Affective Valence Judgments more slowly to the more arousing images and also react
more slowly to new unpleasant pictures than to new
A Dimensional analysis of EXPRESSIVE PHYSIOLOGY: pleasant pictures. The valence difference in reaction time
Facial muscle, heart rate, and sweat gland activity is not found, however, for previously viewed pictures. For
"old" pictures, only arousal is pertinent, and this time
1.5 rraiiin i Tujniniffr rnrr • the fastest reactions are to the more exciting images,
rm.St whether pleasant or unpleasant (Bradley, Greenwald, Pe-
& 1.0
try, & Lang, 1992). Furthermore, when normal subjects
eauo are placed in a free-viewing context (see "control subjects"
I .5 in Figure 3), they tend to spend as much time looking at
3 unpleasant as pleasant pictures. That is, as one might
10 infer from the popularity of "slasher" movies, or the ha-
bitual slowing of traffic at a roadside accident, viewing
-.5 time in normal subjects is correlated with arousal or in-
5 10 15 20 25 5 10 15 20 25
Ratad Pleasure
tensity of evoked emotion rather than pleasantness. Stud-
Bated Pleasure
ies of persons with specific fears, however, suggest that
this does not follow if pictures evoke very high levels of
distress. When phobics viewed pictures of their own pho-
bic objects, a palpable reduction in viewing time was
found, consistent with their general avoidant behavior
pattern (see "animal phobics," Figure 3).
As the fear data implies, relationships between specific
measures can vary widely for individuals and to some
extent between particular groups. For example, pleas-
antness ratings covary more closely with facial muscle
activity in females than in males; on the other hand, skin
conductance changes are more closely correlated with
arousal ratings in males than in females. Overall, however,
the motivational variables predominate in organizing the
Note. Affective valence judgments covaried with slide response {upper left), zyg- picture perception data. Thus, a factor analysis of various
omatic response (upper middle), peak heart rate acceleration (upper right!, and
covariation of arousal judgments with skin conductance response (SCR) magnitude
(lower left). In each case, judgments are rank ordered for each individual; the
graphs depict mean responses at each rank across individuals.(EMG = electro-
myograph; HR = heart rate,- bpm = beats per minute (see Lang et ol., 1993I.
Figure 3
Mean Duration of Free-Viewing for Four Picture Types
Picture viewing time
wald, Cook, & Lang, 1989; Lang, Greenwald, Bradley, & Pleasant
Hamm, 1993). There is little change in the distribution Neutral
of values from initial standardization to retest in exper- Unpleasant
iment, even with different methods of rating administra- Phobic
tion (e.g., paper-and-pencil form vs. subject-controlled
computer graphic). Furthermore, physiological and overt
behaviors in emotion (the other two elements in a three-
systems analysis) are modulated in roughly parallel ways
over picture contents. Thus, as can be seen in Figure 2,
physiological events covary significantly with the param-
eters of valence and arousal, as denned by evaluative
judgments: For a representative sample of I APS pictures,
mean facial muscle activity during viewing correlates
strongly with subjects' mean valence ratings (ranked un-
pleasant to pleasant by picture). Heart rate shows a similar
Animal Phobics Control Subjects
but more modest relationship. On the other hand, mean
skin conductance covaries only with judged arousal, in- Note. Pictures included those judged to be lal highly pleasant, (b) neutral in affect,
Ic) highly unpleasant by the standardization sample, and (d) consisting of snakes
creasing monotonically over the stimulus range. and spiders. Data are presented seporately for normal control subjects and for
Performance behaviors relevant to emotional picture snake and spider phobics (Hamm, Cuthbert, Globesich, & Vaitl, in press).

perception also covary with motivational parameters.

May 1995 • American Psychologist 375

Table 1 circuit, with its efferent output paths, is presented in Fig-
Sorted Factor Loadings of Emotion Measures ure 4.
on Principal Components Input normally passes from the sense organs to the
sensory cortex, although simple sensory information (e.g.,
Measure Factor 1: valence Factor 2: arousal
lights or tones) may require only thalamic processing (e.g.,
Valence ratings
see the cortical lesioning studies of DiCara, Braun, &
.86 -.00
Corrugator muscle -.85 .19 Pappas, 1970). From the sensory-specific nuclei of the
Peak heart rate .79 -.14 thalamus the circuit proceeds to the amygdala—first to
Zygomatic muscle .58 .29 its lateral and basolateral then to its central nucleus
Arousal ratings .15 .83 (LeDoux, 1990). The bilateral amygdala, located within
Interest ratings .45 .77 the temporal lobes of the brain, has long been known as
Viewing time -.27 .76 a critical structure in the mediation of emotional expres-
Skin conductance -.37 .74 sion (see the edited volume by Aggleton, 1992). That is,
Note. The factors having the highest loadings for each variable are highlighted
both stimulation and ablation of this site has reliably al-
in bold.

Figure 4
Schematic Representation of o Part of the Neuronal
affect self-report, physiological, and behavioral measures Network Involved in the Transformation of an Insignificant
resulted in a strong two-factor solution, with pleasantness Acoustic Tone Into a Conditioned Aversive Input
ratings, heart rate, and facial muscles loading on one fac- ,,__._„ lesions of sensory
tor (valence) and arousal and interest ratings, viewing
^STIMULUS
time, and skin conductance all loading highly on a second INPUT
factor (arousal). The cross-loadings for all measures (see
Table 1) are very low. Affects are built around motiva-
tional determinants.
lesions of the lateral
Motive Systems in the Brain hypothatamus block
only the autonomic
response in tear
As proposed at the outset, the structural foundation of conditioning
these valence and arousal effects is in the brain's appetitive
and aversive motive systems. These systems consist of
neurophysiological circuits in the brain, largely subcor-
lesions of the
tical,' that are primitively activated by primary rein- ventral gray
forcement. The present discussion focuses on studies of attenuate HYPOTHALAMUS
"freezing';
the aversive system, in part because of its pertinence to dorsal lesions
enhance It
psychopathology and in part because the relevant database
for this system is larger and the results clearer.
AUTONOMIC
In the main, what is known about the brain's aversive
NERVOUS SYSTEM
system comes from animal research, particularly from
studies of defense reactions and fear conditioning in the
rat. In this species, as in man, massive defensive reactions
(autonomic and behavioral) are readily evoked by painful
BEHAVIORAL BLOOD PRESSURE
stimuli (e.g., electric shock). If such stimuli reliably occur "FREEZING" INCREASE
contiguous with or immediately following a previously
innocuous stimulus (e.g., an auditory tone), features of RESPONSE OUTPUT
the defensive behavior come to be evoked by the formerly Note. The conditional stimulus is transmitted to the acoustic thalamus (through the
innocuous stimulus, presented alone. That is, the new auditory system and shell regions of the inferior colliculus [not shown]). The critical
stimulus comes to activate a pattern of emotional re- thalamic neurons are in the medial geniculute body of the nearby posterior intra-
laminar nucleus. "The lateral nucleus of the amygdala receives the acoustic signal
sponses we call fear. from the thalamus and transmits to the centra! amygdala. Efferent to the amygdala
Investigators are beginning to understand the neural the pathway bifurcates. Projections from the central amygdala to the lateral hy-
pothalamic area mediate the autonomic conditioned emotional response, whereas
basis of aversive reactivity and associated conditioning projections to the mid-brain central gray region mediate the behavioral conditioned
phenomenon. That is, using neurosurgical, pharmaco- response" (LeDoux, 1990, p. 311. Details of the circuit through dorsal and ventral
central gray are given by Fanselow ef al. (1995).
logical, and electrophysiological tools, the chain of prob-
able neural activation can be traced, starting from the
input end in the sensory system—proceeding through
the necessary connecting structures, defining the links 1
These circuits may also involve a latcralized cortical component
least prodigal in synaptic connections—to the autonomic (e.g., Davidson, 1993; Heller, 1993), although the specific, connecting
and motor effectors. An adumbrated conditioned fear neurophysiology has yet to be denned.

376 May 1995 • American Psychologist

tered a variety of affective/motivational behaviors in both fensive immobility (i.e., freezing and hypervigilance in
animals and man (e.g., Aggleton & Mishkin, 1986; Everitt which the organism is passive but "primed" to respond
& Robbins, 1992; Ursin, Jellestad, & Cabrera, 1981). to aversive stimulation).
Furthermore, some evidence suggests that the amygdala
may have greater relevance for negative affect and aver- Motivational Priming
sion-driven than appetitive behaviors—particularly those Scientists and laypeople agree that the emotions loom
controlled by secondary reinforcers (Cahill & McGaugh, large in life. They have, of course, significance as states
1990). in themselves, with their own surgent response patterns.
Efferent to the central amygdala, the aversion circuit In addition, they are held to play an important modu-
branches, with each path apparently governing separate latory role on other behaviors—coloring, inhibiting, or
response outputs. Lesioning studies show that different energizing thought, perception, and action—often in
somatic responses, such as "freezing" (immobility in the contexts that are physically and semantically remote from
face of threat) and active fight or flight, and autonomically the apparent source of the controlling affect. It is proposed
driven increases in blood pressure are mediated through here that this ubiquitous spread of affection across the
different neural centers. The autonomic response is de- behavioral repertory is motivationally mediated.
pendent on an intact pathway through the lateral hypo-
Emotions reflect the engagement of neural structures
thalamus (LeDoux, 1990), and the somatic components
and pathways in either the appetitive or aversive moti-
require an intact midbrain (periaqueductal) central gray
vation systems (e.g., as discussed above). During the pe-
area (Figure 4). Furthermore, the ventral central gray is
riod when this subcortical circuitry is active, a modulatory
the fear freezing path, whereas the dorsal gray is a critical
effect is exerted on the brain's other processing operations.
part of the fight/flight action circuit (see the data and
Specifically, associations, representations, and action
discussion of Fanselow, DeCoIa, De Oca, and Landeira-
programs that are linked to the engaged motivational sys-
Fernandez, 1995, and Depaulis & Bandler, 1991).
tem have a higher probability of access (with a concom-
It is reasonable to assume neural plasticity in these itantly greater potential output strength) than other in-
subcortical circuits. That is, although some subamygda- formation, and conversely, mental events and programs
loid connections might be "hard wired," the brain linked to the nonengaged system have a reduced proba-
"learns" at many levels of its anatomy and functioning. bility and strength of activation. Thus, in the case of an
Depending on variations in the built in and acquired aversively motivated organism (i.e., the affective state is
weighting of these local pathways, considerable variety in unpleasant), responses to other aversive cues are primed,
the pattern of affective output could be expected. In fact, and at the same time responses to appetitive cues may be
the amygdala has been implicated in many aversively reduced or absent. As an example, recent findings (Blaney,
motivated responses, that is, in escape and avoidance 1986; Bower, 1981) suggesting that negative or positive
learning (Ursin, 1965); in defensive/aggressive behavior affective moods occasion emotionally congruent verbal
(Blanchard & Blanchard, 1977; Roldan, Alvarez-Pelaez, associations and memories might be construed to be a
& Fernandez de Molina, 1974); and, as is examined in result of such motivational priming.
more detail later, in augmenting startle reactions (e.g.,
Davis, 1989). Furthermore, this same aversive system also The most primitive and fundamental motivational
prompts varied hormonal and autonomic responses. For priming is, however, at the level of unconditioned extero-
example, Iwata and LeDoux (1988) noted that heart rate ceptive reflexes. Recall that reflexes can be sorted (Ko-
and blood pressure decrease in response to a conditioned norski, 1967) according to the primary reinforcement
tone (previously accompanied by shock) when animals properties of their unconditioned stimuli, either appetitive
are physically restrained; on the other hand, these same or aversive, and the consummatory or defensive function
autonomic responses increase when the conditioned signal of reflex itself. In the priming view, responses to uncon-
is presented to freely behaving animals. These data recall ditioned stimuli are modulated according to two factors:
similarly diverse context effects in humans, who show (a) the classification of the reflex (appetitive or defensive)
heart rate acceleration while imaging or thinking about and (b) the affective valence of the individual's ongoing
unpleasant events (e.g., Vrana & Lang, 1990) but show emotional state. Thus, an independently evoked defensive
significant deceleration when shown unpleasant pictures reflex will be augmented when the organism is already
(e.g., Lang et al., 1993). reacting to an aversive foreground stimulus (i.e., is in an
unpleasant state); this same reflex will be reduced in am-
Overall, the neurophysiological findings suggest that plitude when the organism is processing an appetitive
the amygdala is a general mediator of defensive behaviors, foreground. Finally, both these priming effects—poten-
for example, a key site in a general aversive motivational tiation and diminution of responding—are expected to
system. Structures downstream from the amygdala are be enhanced according to the level of affective drive or
implicated in the different forms of defense. Although activation.
context dictates that these autonomic and somatic pat-
terns have great variety, they can be functionally organized Startle Potentiation
into two general classes: (a) defensive action, (i.e., con-
textual variations in fight/flight that are more or less direct In most mammals, any abrupt sensory event will prompt
responses to nociception or imminent attack) and (b) de- a chained series of rapid flexor movements that can cas-

May 1995 • American Psychologist 377

cade throughout the body (Landis & Hunt, 1939). This
startle response appears to be a primitive defensive reflex Figure 5
that serves a protective function, avoiding organ injury Schematic Representation of the Primary Neural Path
(as in the eyeblink) and acting as a behavioral interrupt Between o Startle Stimulus Input and Its Effector Output
(Graham, 1979) that clears processors to deal with pos- SHOCK SENSITIZATION
sible threat.
SHOCK
According to the priming hypothesis, the defensive electrical stimulation
CONDITIONING of the amygdala enhances
startle reflex should be considerably faster and of signif- the acoustic startle reflex
icantly greater amplitude when the aversive motivational
system is already active, for example, as in a fear state. lesions of the
amygdala abolish BASOALATERAL,
CENTRAL
This was first examined systematically by Brown, Kalish, potentiated startle & LATERAL
NUOEUS
NUCLEUS
and Farber (1951), who compared reflex responses with
lesions in different regions
startle probes (shots from a toy pistol) presented to male AMYGDALA of the central gray may
rats during neutral or shock conditioned stimuli at ex- facilitate or abolish
potentiated startle
tinction. Results conformed to expectation: Animals did
indeed react more forcefully—as measured by a stabili-
electrical startle
meter in the floor of the cage—when the startle stimuli stimulation at this
were presented during fear-conditioned signals (see also site: fear potentiation
electrical startle
is still observed
Ross, 1961; Spence & Runquist, 1958). stimulation at this
site: Fear potentiation
NUCLEUS RETICULARIS is not observed
The Rat Brain's Fear-Startle Circuit PONTIS CAUDALIS
Davis and his associates (e.g., Davis, 1989; Davis, Hitch-
cock, & Rosen, 1987; Lee et al., 1994) and others (Fendt, \
Koch, & Schnitzler, 1994a) have since gathered consid- INPUT PATH SPINAL* X OUTPUT PATH
erable evidence that the brain structure mediating fear- [ FACIAL MOTO- ]
.NEURONS
conditioned startle potentiation is, at least in the rat, the
same aversive system that was previously described. A
diagram of the relevant neural paths is presented in Figure
ABRUPT NOISE STARTLE REFLEX
5: Beginning with stimulation of the ear by an acoustic
probe (e.g., abrupt noise), the afferent path of the startle Note. Locations along this path are shown where stimulation with microelectrodes
will, or will not, produce a potentiated startle response. This procedure isolated
reflex proceeds from the cochlear root neurons to the the neuronal structure (nucleus reticuloris pontis caudolis) where the modulating
reticular nucleus; efferent connections pass from there influence of previous oversive learning impacts on the obligatory startle circuit.
through spinal neurons to the reflex effectors. This is the The central nucleus of the amygdala Ithe same central structure as in LeDoux's,
1990, description of an aversive learning network) projects directly to this reticular
basic obligatory circuit, directly driven by the parameters site, and through the central gray. The fact that lesioning of the amygdala blocks
of the input stimulus (e.g., stimulus intensity, frequency, potentiation, and stimulation with microelectrodes potentiates the reflex (in the
absence of prior fear conditioning), prompts the conclusion that this structure is o
and steepness of the onset ramp). primary component of the oversive motivational system (see Davis, 1986, 1989,
The phenomenon of conditioned startle potentiation in press, for an overview of the neurophysiology).
implies, however, a secondary circuit that modulates this
primary reflex pathway. There is now overwhelming ev-
idence that the amygdala, the key structure in aversively
motivated behavior, is a critical part of this modulatory
to be disrupted by lesions in the dorsal pathways (e.g.,
circuit. It has been shown that, first, there are direct,
Fendt, Koch, & Schnitzler, 1994b). Interestingly, when
monosynaptic projections from the amygdala to the key
the grosser defense responses (fight or flight) are engaged
reticular site, that is, to the structure in the basic circuit
they appear to activate the dorsal central gray (Fanselow
on which modulation of the reflex depends; second, elec-
et al., 1992) and interfere with the expected increase in
trical stimulation of the amygdala (below the level for
potentiated startle. In fact, lesions of the dorsal central
kindling) directly enhances startle reflex amplitude; and
gray can restore potentiated startle in some circumstances
finally, and most significant, lesions of the amygdala
(D. Walker & M. Davis, personal communication, March,
abolish fear-conditioned startle potentiation.
1995).
It is important to note that startle potentiation is cor-
related with freezing behavior in the rat. Thus, for ex- Human Conditioning and the Blink Response
ample, foot shocks presented in a destructive environment
will prompt both freezing (Blanchard, Fukunaga, & Fear conditioning can be demonstrated in humans, of
Blanchard, 1976) and potentiated startle (Davis, 1989). course, using a visual stimulus-shock paradigm similar
Passive freezing (Fanselow et al., 1992), and perhaps po- to that used in studies of the rat. Furthermore, the phe-
tentiated startle as well (Fendt, Koch, & Schnitzler, nomenon of startle reflex potentiation can also be ad-
1994b), are disrupted by lesions in the ventral periaq- dressed in humans, although the methodology for re-
ueductal gray (Fanselow et al., 1994). Fear potentiated cording the response is somewhat different. As already
startle, on the other hand, a reflex action, is more likely noted, the whole-body startle reaction is typically re-

378 May 1995 • American Psychologist

corded in animal studies. In research with humans, in- a passive vigilance task in which—like a freezing rat or an
vestigators usually measure only the eyeblink. attentive predator—viewers are generally immobile. When
Rapid eye closure is one of the first and most reliable startle probes are administered in this context, results have
components of the behavioral cascade that constitutes consistently conformed to the motivational priming hy-
the startle reflex. Occurring 30 to 40 milliseconds after pothesis: A significant linear trend is reliably observed over
stimulus onset, it reflects an abrupt increase in tension judged picture valence, indicating that the largest startle
in orbicularis oculi—the facial muscle that surrounds the blink responses occur during unpleasant content and the
eye. The latency and magnitude of the blink can be mea- smallest during pleasant pictures (see Bradley, Cuthbert, &
sured by monitoring this muscle, using miniature elec- Lang, 1990; Greenwald, Bradley, Cuthbert & Lang, 1990;
trodes placed just beneath the lower lid. Vrana, Spence, & Lang, 1988).
In studies with humans, the stimulus used to evoke Bradley, Cuthbert, and Lang (1991, in press) found this
the blink is relatively modest—typically a 50-millisecond affect-reflex effect using monaural rather than binaural
burst of white noise at around 95 decibels, which, al- acoustic probes and furthermore obtained evidence that the
though it prompts a clear blink response, rarely interferes effect may be lateralized (i.e., left ear probes, presumably
with ongoing foreground tasks. Several researchers have conferring an advantage in right brain processing, showed
confirmed potentiation of the blink response in human the strongest relationship with affective valence). Jansen and
aversive sensitization or conditioning that parallels startle Fridja (1994), using evocative video film clips, and Hamm,
studies of the rat (Greenwald, Bradley, Cuthbert, & Lang, Stark, and Vaitl (1990), using IAPS slides, obtained the af-
1991; Hamm, Greenwald, Bradley, & Lang, 1993). In fect-startle effect in European research participants. Finally,
brief, the blink muscle response to a startle probe is gen- Bradley et al. (1990) found the same pattern of affective
erally larger after subjects experience electric shock and modulation using both visual and acoustic startle probes
selectively larger to startle probes presented during ex- (see Figure 6), disconfirming an alternative hypothesis that
posure to a shock conditioned stimulus than to probes affective modulation is secondary to differences in modality-
presented during exposure to an unshocked control stim- directed attention (Anthony & Graham, 1985).
ulus. These results, coupled with human neurological ev- More recent results have further extended and clarified
idence linking the amygdala to aversive emotion (e.g., this phenomenon. On the one hand, it has been shown
Morris, Bradley, Bowers, Lang, & Heilman, 1991), en- that over many presentation trials of the same picture
courage the hypothesis that similar neural pathways might stimuli, despite a marked general habituation of the startle
be responsible for potentiation effects in both rats and reflex itself, affective potentiation and inhibition persist
humans. In addition, the data suggest that the startle (Bradley, Lang, & Cuthbert, 1993). On the other hand,
probe methodology might be a valuable tool for assessing it is now apparent that the general affect-startle effect is
the basic view presented here; that is, the aversive and more pronounced in the context of more arousing stimuli.
appetitive motivational systems play a fundamental role Cuthbert, Bradley, and Lang (in press) showed that when
in organizing human emotional expression. unpleasant slides are rated highest in arousal, both skin
conductance and probe startle potentiation are greatest.
Probing Emotional Perception Similarly, as pleasant pictures are judged to be more
It is proposed that the startle reflex is differentially mod- arousing, skin conductance (indexing sympathetic acti-
ulated depending on an individual's ongoing emotional vation) increases, as for unpleasant pictures. For startle
state. More specifically, the amplitudes of blinks evoked magnitude, however, increasing arousal of a pleasant
by startle probes will vary with the valence of the affective foreground prompts an opposite effect—greater reflex in-
state prompted by a foreground percept. In an unpleasant hibition. This result is consistent with the motivational
stimulus foreground the brain is processing negative af- priming hypothesis proposed earlier. Arousal does not
fective information; the relevant subcortical, aversive sys- have an independent, unitary effect on behavior: Mod-
tem circuitry is contacted; and a defensive reflex is aug- ulation of the defensive reflex increases with activation,
mented. Appetitive and aversive/defensive dispositions are but the direction of effect is different, depending on the
considered to be opponent states. That is, the active dis- motivational system (appetitive or aversive) engaged.
position (appetitive or aversive) controls the output pro- Other data support the hypothesis that the modulatory
cessors, and affectively nonconcordant reflexes will not pattern described here is specific to defense-type reflexes.
be potentiated and could actually be inhibited. Thus, For example, a T-reflex probe (activation of the soleus mus-
when appetitive information is the focus of processing, cle in the leg with a tap to the Achilles tendon) does not
the startle reflex is expected to show relative diminution. prompt a differential modulation by picture valence. Instead,
Direct tests of this hypothesis have used stimuli from the for this neutral posture reflex, a simple arousal effect is
IAPS (Center for the Study of Emotion and Attention, 1994), observed, with larger responses to both arousing pleasant
selected on the basis of normative affective ratings and or- and unpleasant stimuli relative to the calmer judged neutral
ganized into three affective classes—unpleasant (e.g., poi- content (Bonnet, Bradley, Lang, & Requin, in press). Fur-
sonous snakes, aimed guns, and pictures of violent death), thermore, it appears that responses indexing the brain's at-
pleasant (e.g., happy babies, appetizing food, and attractive tentional system—that are not motivationally specific—also
nudes), and neutral (e.g., umbrellas, hair dryers, and other do not show modulation according to affect system. Thus,
common household objects). Picture viewing is, of course, the P300 component of the cortical event-related potential,

May 1995 • American Psychologist 379

 pictures. Psychopaths did not differ from students or other
Figure 6 prisoners in their affective ratings of these pictures, judging
Mean Blink Reflex Magnitudes to Acoutic and Visual them to be arousing and unpleasant. Such response dis-
Startle Probes cordance is consistent with the psychopathic diagnosis.

n1
Acoustic Startle Probe In fact, a further analysis of the data showed that the
1075 -f potentiation deficit was not related to criminality per se
Pleasant but was observed specifically in prisoners who, at inter-
• Neutral
Unpleasant
view, showed the characteristic emotional indifference and
lack of remorse of the classic psychopath (Cleckley, 1976).
1025- • As would be predicted, phobic individuals exhibit

1
greater startle potentiation while viewing pictures of their
own phobic objects than do nonphobic individuals view-
ing these or other unpleasant pictures (Hamm, Cuthbert,
975- Globisch, & Vaitl, in press). It does not appear to be the
case, however, that all anxiety states show the same pattern
of reflex response. Greater potentiation appears to occur
most reliably with fears of specific external pain or harm.
095 - S

Visual Startle Probe

500

Figure 7
Imagery and Startle Probe Magnitude

350

Note. Reflex magnitudes are to acoustic (top panel) and visual (bottom panell
startle probes presented during viewing of pleasant, neutral, and unpleasant pic-
torial slides (Bradley, Cuthbert, & Lang, 19901. IBoth the acoustic and visual startle 47
probes resulted in blink magnitudes with similar linear patterns over valence cat- • NEUTRAL 9 SOCIAL 0 DANGER • CLINICAL
egories [pleasant < neutral < unpleasant], A/D — analog-to-digital conversion). IMAGERY CONTENT
Adapted from "Emotion, Attention, and the Startle Reflex" by P. J. Lang, M. M.
Bradley, and B. N. Cuthbert, 1990, Psychological Review, 97, p. 384. Copyright
1990 by the American Psychological Association.

recorded in response to startle probes, shows only an arousal

effect, that is, a larger wave form during interesting, affective
stimuli than during neutral pictures (Schupp, Cuthbert,
Bradley, Lang, & Birbaumer, 1993).

Individual Differences and 48

Psychopathology SIMPLE PTSD SOCIAL
PATIENT DIAGNOSIS
The affect-startle effect has an impressive ubiquity. A
Note. As part of an initial assessment for treatment, four groups of anxiety patients
recent study by Balaban (1995) even found it in five- (overall N = 90) were instructed to image the content of specific sentences, involving
month-old infants, whose blink responses to startle probes either routine and affectively neutral events, anxiety evoking social situations (an
were larger when viewing pictures of angry than when embarrassing act, a speech performance), situations involving threat of danger
(an auto accident or a night time intruder), or situations directly relevant to their
viewing happy faces. An interesting failure to observe the specific clinical problem. All patients showed larger startle reflexes to probes
phenomenon, however, occurred in a picture study of presented during fear than during neutral sentence imagery. Overall mean probe
startle reflex magnitudes to the different fear imagery contents are shown in the
incarcerated prisoners (Patrick, Bradley, & Lang, 1993). upper part of the figure. Reflex means for the subgroups—simple phobia, post-
Whereas inmates without psychiatric diagnoses responded traumatic stress disorder (PTSDI, social phobia, and panic disorder—are shown
as did college students, psychopathic prisoners failed to in the lower part of the figure.
show startle reflex potentiation when viewing unpleasant

380 May 1995 • American Psychologist

 Figure 7 shows the blink reflex responses of four types the brain's output processors. However, a stimulus con-
of psychiatric patients, each group with a different anxiety figuration can be multidimensional and doubly evocative.
diagnosis. Probe startle stimuli were presented while the In this case, a conflict theory is assumed (of the sort pro-
patients imaged neutral scenes and three types of specific posed many years ago by Neal Miller [1944]) in which
fear scenes: (a) events involving their own specific clinical the behavioral resolution is determined by the relative
concern, (b) events that would be dangerous or painful strength (activation level) of each motive state.
experiences for anyone (e.g., a night intruder when home Certain affects, indeed, seem to intrinsically involve
alone or an automobile accident), and (c) events that in- dual motivation. In Figure 8 probe startle reactions are
volved a difficult or embarrassing social performance. shown to picture stimuli that have been rated by an in-
Overall, the patients' startle reflexes during fear imagery dependent sample on separate scales for 10 experienced
were largest for scenes that involved potential physical emotions. For pictures receiving the highest relative score
danger and smallest during scenes of social distress. For on sexual pleasure (e.g., an attractive, seminude couple),
both social phobics and panic patients, probe responses the reflex was smallest, as expected; conversely, both dis-
during the danger scenes exceeded responses during gust and fear evoked larger responses, with dominantly
scenes of clinical concern. fearful stimuli showing the strongest reaction of all these
These results support the hypothesis that startle po- picture categories (see also Balaban & Taussig, 1994). In
tentiation reflects aversive system engagement, that is, this example, the pictures receiving the highest rating for
subcortical structures related to harm avoidance, which pity actually exceeded fear pictures in arousal (6.8 vs.
in the rat are activated by primary unconditioned rein- 5.7) and were judged lower in pleasure (1.9 vs. 3.6); nev-
forcers, such as electric shock, and by their associated ertheless, they showed less probe potentiation (see Figure
conditioned stimuli. It is not unexpected that simple 8). Interestingly, the pity pictures (e.g., pictures of babies
phobics and post-traumatic stress disorder patients would undergoing extreme medical treatments or starving peo-
show large reactions to their own clinical scenes (even ple) were also rated high in nurturance—an approach
larger than to the standard danger scenes). For these pho- disposition to help and care for others. The relative startle
bia and trauma diagnoses, the clinical scene is about a attenuation observed for pity pictures might represent
more imminent concern, and (unrealistic though it may the algebraic sum of opponent motives, appetitive and
be) it is also about physical pain and danger (e.g., threat-
ening animals, falling, blood and injury, assault, rape, or
combat). Conversely, social anxieties and the often vague
worries and uncertainties found in many panic patients Figure 8
may be more consistent with inhibitory states, as de- Mean Blink Reflex Magnitudes fo Startle Probes
scribed by Gray (1985), and less an expression of active,
goal-specific escape and avoidance motivation. These data
Reported feelings and startle reflexes
suggest that the startle paradigm may prove to be of con-
551
siderable value in sharpening diagnosis of anxiety states
and in relating psychopathology to functional brain ac-
tivity.

Conclusions
Research Paths and Detours
The reader has certainly already discerned the limitations
of the model of emotion presented here. For example,
the focus has been on action dispositions—on what Wil-
liam James (1894) called the "coarse" affects. This article
has not considered more subtle "feelings" or the general
factors that block or delay action and thus create the dis-
positional set. Clearly, the latter factors are of great variety
and determine the emotions' varied tactical shapes. They
include immediate external factors, for example, the size
of an aggressor may be of importance in determining 40
whether an aversive affect is fear (implicit flight) or anger Sexual pleasure Disgust Fear
(attack), as well as internal factors that go beyond the
Reported Emotion
current discourse, such as health, biological temperament,
and learning. Note. Mean blink reflex magnitudes to startle probes presented while individuals
Dual motives. A further complication may arise viewed pictures that evoked feelings of sexual pleasure, pity, disgust, and fear.
Participants rated a large sample of pictures on o group of basic emotion scales.
when a stimulus context simultaneously activates both The mean reflex values shown here are for pictures that received consistently high
motive systems. I have suggested that the appetitive (ap- ratings for a specific affect 13-9 pictures per emotion!.
proach) and aversive (withdrawal) systems compete for

May 1995 • American Psychologist 381

aversive—a phenomenon that could be obscured by bi- modality and directional tuning of perceptual processors,
polar affective judgments. motor inhibition, and reduced cardiac effort.4 According
Anticipation, alarm, and helplessness. The to Sokolov's (1963) classic analysis, perceptual processing
dispositional nature of affects imposes another limitation is activated by any change in stimulus input. Thus, each
on startle analysis. Recall that potentiated startle in the picture presentation should prompt phasic orienting,
animal model is found most reliably in fearful, immobile varying in strength, dependent on stimulus parameters.
individuals, not in individuals actively avoiding or escap- Basing his view on studies of simple shocks, tones, and
ing. The view presented here is similar to that of Mas- noise, Sokolov (1963) emphasized changes that occur with
terson and Crawford (1982). They hypothesized a "de- an increase in physical intensity. In his view, as the loud-
fense motivation system" in animals that "selectively po- ness of an input augments, orienting first increases and
tentiates or primes a set of innate reactions that include then gives way to a new reaction, the defense response—
fleeing, freezing, fighting and defensive burying" (p. 664). at first on only some trials and then as a consistent reaction
They conceived the human experiences of fear, anger, to stimulation. Defense responses reflect a suspension of
anxiety, and apprehension to be evolved expressions of processing—input rejection—and augur active escape.
this system. Furthermore, these authors made an impor- High auditory intensity is, of course, both aversive and
tant distinction between priming of the defense motiva- arousing. If one applies the Sokolovian model to startle
tion system and its behavioral sequelae—the "alarm re- probe data gathered during emotional pictures, and as-
action." When the organism has been defensively dis- sumes that intensity and judged arousal are analogous
posed, related reflexes are pre-set: "An excited defense parameters, there are striking parallels.
response is like a cocked pistol ready to be triggered by Frances Graham and others (e.g., Anthony & Graham
supporting stimuli" (p. 664). It is important to note, 1985, Simons & Zelson, 1985) hold that probe startle
however, that the "alarm reaction" is not part of this inhibition indexes attention (i.e., interesting foregrounds
preparatory set. It is a possible sequela of overt action— use attentional resources thus less are available for probe
fight or flight—for which the organism has been mobi- processing, resulting in reflex inhibition). As can be seen
lized. As noted by Lang, Bradley, and Cuthbert (1990), in Figure 9, as pleasant pictures are judged more arousing
startle potentiation is to be observed during this first, (and interesting) inhibition does increase. Furthermore,
"cocked pistol," stage. After the alarm stage has been unpleasant stimuli appear to show a similar coincidence
triggered and the organism swings into action—actively of increased picture arousal and startle inhibition—at
coping or in headlong panic—the harm-anticipatory de- least in the mid-range. As with Sokolov's (1963) defense
fense system is overwhelmed. Part of potentiated startle's response, however, this effect is reversed for unpleasant
neural path is preempted by action,2 and blinks no longer stimuli that are farther along the arousal continuum. In-
reflect affective subtleties. deed, at a certain point, a monotonic increase in startle
A further constraint on the affect-startle effect involves potentiation is observed.
states of the organism characterized by somatic or au- These findings highlight important theoretical issues
tonomic inhibition. For example, startle potentiation is and signal important directions for research. First, re-
attentuated by anxiolytic drugs in both animals (Davis, searchers need to know much more about the covariation,
1986) and humans (Patrick, Berthot, & Moore, 1993). and possible interactions, of the various measures of ori-
Interestingly, whereas a moderate amount of alcohol re-
duces overall startle amplitude, it does not prevent emo- 2
Potentiated startle may compete with larger, somatically mediated
tional modulation (Stritzke, Patrick, & Lang, 1995). Also, actions for the pathway from the amygdala through the dorsal central
startle analysis might not inform, if the individual's psy- gray area, suggested by lesion studies (M. Davis, personal communication,
chological state precluded attention and action. This may October 28, 1994).
3
be true for affects such as depression. In animals, long Weiss (1989) has suggested that the inactivity that characterizes
depression may be a consequence of norepinepherine depletion in patients
periods of punishment can produce "learned helpless- with this diagnosis.
ness" (Seligman, 1975). The organism is no longer dis- 4
"People look at photographs because it is pleasant and interesting
posed to act—depleted by a persistent stressful context to do so. . . . Observers are invited to attend to such stimuli, but it is
that permits no behavioral recourse. This may also char- generally understood that most instrumental behavior is to be inhibited
(or closely regulated) according to learned aesthetic rules. In lay terms,
acterize states in which important positive reinforcers perception occurs across an aesthetic distance, denned by the 'artistic
have been irrevocably removed, as in grief after great per- frame of reference . . . around the work . . . to differentiate it psycho-
sonal loss.3 In these cases, a state of high negative valence logically from reality'" (Gove, 1986, p. 34). "Nevertheless, pictorial in-
exists, but active scanning of the environment is absent. formation can match the stimulus properties of real object or event
referents, activating cognitive representations associated with strong
In this circumstance there may be a reduction in reflex emotional responses. Affective processing, in turn, triggers facial action
response and little evidence of startle modulation. and visceral motility similar in pattern to that engaged by the veritable
stimuli, which can closely parallel evaluations of affective meaning. Be-
Stop! Look! Listen! cause the individual's aesthetic set generally assures that overt actions
(e.g., flight or aggression) are gated out of emotional response programs,
As the above suggests, affective startle modulation occurs residual affective behaviors in the aesthetic context are those related to
most reliably during states of vigilance, when the organ- initiation and choice—assuming that selection of stimuli is freely avail-
ism is stopped but actively orienting. Picture viewing is able—and the active maintenance or termination of stimulus input"
quintessential^ such an orienting task, characterized by (Lang, Greenwald, Bradley, & Hamm, 1993, p. 262).

382 May 1995 • American Psychologist

 area. Furthermore, the demonstrated replicability of the
basic human research findings, within and across labo-
Figure 9
Changes in Probe Startle Magnitude
ratories, encourages the belief that we in the emotion
world have at last reached a kind of terra firma. Mindful
0.4 T
that the complexities are great and that discoveries often
confound cherished preconceptions, we are eager to carry
•§
N 0.2 forward this search for new solutions to the old mysteries
•s of human feeling.
eg
•o
« 0.0 REFERENCES

Aggleton, J. P. (1992). The Amygdala. New York: Wiley.

-0.2 Aggleton, J. P., & Mishkin, M. (1986). The amydgala: Sensory gateway
to the emotions. In R. Plutchik & H. Kellerman (Eds.), Emotion:
I -0.4
Theory, research and experience (Vol. 3). New York: Academic Press.
Anthony, B. J., & Graham, F. (1985). Blink reflex modification by se-
Pleasant lective attention: Evidence for the modulation of "automatic" pro-
c
m Unpleasant cessing. Biological Psychology, 21, 43-59.
Balaban, M. T. (1995). Affective influences on startle in five-month-old
-0.6
5.0 5.5 6.0 6.5 7.0 7.5 infants: Reactions to facial expressions of emotion. Child Development,
66, 28-36.
Normative Arousal Rating Balaban, M. X, & Taussig, H. N. (1994). Salience of fear/threat in the
Nole. Thefigureshows changes in probe startle magnitude separately for pleasant affective modulation of the human startle blink. Biological Psychology.
and unpleasant pictures, as pictures are judged to increase in arousal (IAPS nor- 38, 117-132.
mative valence and arousal ratings!. From 38 to 47 subjects contribute to each Blanchard, R. J., & Blanchard, D. C. (1977). Aggression behavior in
data point, due to rejected trials and the fact that subjects were not startled during the rat. Behavioral Biology, 21, 197-224.
all slides. For clarity, the data were smoothed prior to graphing with a five-point Blanchard, R. J., Fukunaga, K. K., & Blanchard, D. C. (1976). Envi-
moving average (Rafferty & Norling, 1989). ronmental control of defensive reactions to footshock. Bulletin of the
Psychonomic Society, 8, 129-130.
Blaney, P. H. (1986). Affect and memory: A review. Psychological Bul-
letin, 99, 229-246.
Bonnet, M., Bradley, M. M., Lang, P. J., & Requin, J. (in press). Mod-
enting and defense (e.g., as denned by Lacey, 1959; So- ulation of spinal reflexes: Arousal, pleasure, action. Psychophysiology.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36,
kolov, 1963). For example, long after the point on the 129-148.
arousal continuum where defensive startle potentiation Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1990). Startle reflex
begins, heart rate continues to show the decelerative re- modification: Emotion or attention? Psychophysiology, 27, 513-523.
sponse associated with orienting (Graham & Cliffton, Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1991). Startle and emo-
tion: Lateral acoustic stimuli and the bilateral blink. Psychophysiology,
1966). Furthermore, it is now increasingly important to 28, 285-295.
develop a good neurophysiological model of the brain's Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (in press). Lateralized
appetitive system and determine what pathways prompt startle probes in the study of emotion. Psychophysiology.
startle inhibition and how they relate to basic attentional Bradley, M. M., Greenwald, M. K., Petry, M., & Lang, P. J. (1992).
mechanisms. Is the mechanism of inhibition in attention Remembering pictures: Pleasure and arousal in memory. Journal of
Experimental Psychology: learning, Memory, & Cognition, 18, 379-
the same as the hypothesized reduction by appetitive mo- 390.
tivation? That is as yet unknown (e.g., see Swerdlow, Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: The self-
Caine, Braff, & Geyer, 1992, p. 186). assessment manikin and the semantic differential. Journal of Behavior
Finally, it is important to know if the phenomena seen Therapy and Experimental Psychiatry, 25, 49-59.
Bradley, M. M., Lang, P. J., & Cuthbert, B. N. (1991). The Gainesville
in picture viewing can be as consistently observed with murders: Imagining the worst [Abstract]. Psychophysiology, 2S(Suppl.
other modalities of emotional stimulus presentation. Re- 1), S14.
searchers need to know more about responses to moving Bradley, M. M., Lang, P. J., & Cuthbert, B. N. (1993) Startle reflex
pictures, emotional sounds and cries (Bradley, Zack, & habituation in human beings: Emotion, novelty, and context. Behav-
Lang, 1994), and reactions when reading narrative text ioral Neuroscience, 107, 970-980.
Bradley, M. M., Zack, J., & Lang, P. J. (1994). Cries, screams, and
(Spence, Fischler, & Lang, 1994). There are already in- shouts ofjoy: Affective responses to environmental sounds [Abstract].
dications (Bradley, Lang, & Cuthbert, 1991) that reflex Psychophysiology, i/(Suppl. 1), S29.
modulation may not be the same in memory imagery as Brown, J. S., Kalish, H. I., & Farber, I. E. (1951). Conditioned fear as
in perception. More practically, one can ask what are the revealed by magnitude of startle response to an auditory stimulus.
implications of this method for emotion evocation and Journal of Experimental Psychology, 32, 317-328.
Cacioppo, J. X, & Berntson, G. G. (1994). Relationships between atti-
control in clinical treatment and for understanding the tudes and evaluative space: A critical review with emphasis on the
social-developmental effects of media? separability of positive and negative substrates. Psychological Bulletin,
This is a rich terrain for investigation that offers a 115, 401-423.
plethora of exciting paths to follow. Happily, thanks to Cahill, L., & McGaugh, J. L. (1990). Amygdaloid complex lesions dif-
ferentially affect retention of tasks using appetitive and aversive re-
many current and past scholars, researchers have a robust inforcement. Behavioral Neuroscience, 104, 532-543.
theoretical conception, a good neuroscience foundation, Center for the Psychophysiological Study of Emotion and Attention.
and a functional animal model to guide a mapping of the (1994). The International Affective Picture System {photographic

May 1995 • American Psychologist 383

 slides]. Gainesville, FL: Center for Research in Psychophysiology, Hamm, A. O., Greenwald, M. K., Bradley, M. M., & Lang, P. J. (1993).
University of Florida. Emotional learning, hedonic change, and the startle probe. Journal
Cleckley, H. (1976). The mask ofsanity (5th ed.). St. Louis, MO: Mosby. of Abnormal Psychology, 102, 453-465.
Cuthbert, B. N., Bradley, M. M., & Lang, P. J. (in press). Probing picture Hamm, A. Q, Stark, & Vaitl, D. (1990). Classical fear conditioning and
perception: Activation and emotion. Psychophysiology. the startle probe reflex. Psychophysiology, 27, S37.
Davidson, R. J. (1993). Parsing affective space: Perspectives from neu- Hebb, D. O. (1949). The organization of behavior: A neuropsychological
ropsychology and psychophysiology. Neuropsychology, 7, 464-475. theory. New York: Wiley.
Davis, M. (1986). Pharmacological and anatomical analysis of fear con- Heller, W. (1993). Neuropsychological mechanisms of individual differ-
ditioning using the fear-potentiated startle paradigm. Behavioral ences in emotion, personality, and arousal. Neuropsychology, 7, 476-
Neuroscience, 100. 814-824. 489.
Davis, M. (1989). Sensitization of the acoustic startle reflex by footshock. Hess, W. R. (1957). The functional organization of the diencephalon.
Behavioral Neuroscience. 103, 495-503. New York: Grune & Stratton.
Davis, M. (in press). Brain mechanisms involved in potentiation of the Iwata, J., & LeDoux, J. E. (1988). Dissociation of associative and non-
acoustic startle reflex by conditioned fear. In P. J. Lang, M. Balaban, associative concomitants of classical fear conditioning in the freely
& R. Simons (Eds.), Studies of attention. Hillsdale, NJ: Erlbaum. behaving rat. Behavioral Neuroscience, 102, 66-76.
James, W. (1894). The physical basis of emotion. Psychological Review.
Davis, M., Hitchcock, J., & Rosen, J. (1987). Anxiety and the amygdala:
A 516-529.
Pharmacological and anatomical analysis of the fear potentiated startle
Jansen, D. M., & Frijda, N. (1994). Modulation of acoustic startle re-
paradigm. In G. H. Bower (Ed.), Psychology of learning and motivation
sponse by film-induced fear and sexual arousal. Psychophysiology, 31,
(Vol. 21, pp. 263-305). New York: Academic Press.
565-571.
Depaulis, A., & Bandler, R. (1991). The midbrain periaqueductal gray Konorski, J. (1967). lntegrative activity of the brain: An interdisciplinary
matter: Functional, anatomical, and neurochemical organization. New approach. Chicago: University of Chicago Press.
York: Plenum Press. Lacey, J. I. (1958). Psychophysiological approaches to the evaluation of
DiCara, L., Braun, J. J., & Pappas, B. (1970). Classical conditioning psychotherapeutic process and outcome. In E. A. Rubinstein &
and instrumental learning of cardiac and gastrointestinal responses M. B. Parloff(Eds.), Research in psychotherapy (Vol. 1, pp. 160-208).
following the removal of neocortex in the rat. Journal of Comparative Washington, DC: American Psychological Association.
and Physiological Psychology, 73, 208-216. Lacey, J. I., & Lacey, B. C. (1970). Some autonomic-central nervous
Dickinson, A., & Dearing, M. F. (1979). Appetitive-aversive interactions system interrelationships. In P. Black (Ed.), Physiological correlates
and inhibitory processes. In A. Dickinson & R. A. Boakes (Eds.), of emotion (pp. 205-227). New York: Academic Press.
Mechanisms of learning and motivation (pp. 203-231). Hillsdale, NJ: Landis, C , & Hunt, W. A. (1939). The startle pattern. New York: Farrar.
Erlbaum. Lang, P. J. (1968). Fear reduction and fear behavior: Problems in treating
Everitt, B. J., & Robbins, T. W. (1992). Amygdala-ventral striated in- a construct. In J. Schlien (Ed.), Research in psychotherapy, III (pp.
teractions and reward related processes. In J. Aggleton (Ed.), The 90-103). Washington, DC: APA.
Amygdala. New York: Wiley. Lang, P. J. (1980). Behavioral treatment and bio-behavioral assessment:
Fanselow, M. S., DeCola, J. P., De Oca, B. M., & Landeira-Fernandez, Computer applications. In J. B. Sidowski, J. H. Johnson, & T. A.
J. (1995). Ventral and dorsolateral regions of midbrain periaqueductal Williams (Eds.), Technology in mental health care delivery) systems
gray (PAG) control different stages of defensive behavior. Aggressive (pp. 119-137). Norwood, NJ: Ablex.
Behavior. 21, 63-77. Lang, P. J. (1994). The motivational organization of emotion: Affect-
Fendt, M., Koch, M., & Schnitzler, H. (1994a). Amygdaloid noradren- reflex connections. In S. Van Goozen, N. E. Van de Poll, & J. A.
aline is involved in the sensitization of the acoustic startle response Sergeant (Eds.), The Emotions: Essays on emotion theory (pp. 6 1 -
in rats. Pharmacology, Biochemistry, and Behavior, 48, 307-314. 93). Hillsdale, NJ: Erlbaum.
Fendt, M., Koch, M., & Schnitzler, H. (1994b). Lesions of the central Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, attention,
gray block the sensitization of the acoustic startle response in rats. and the startle reflex. Psychological Review, 97, 377-395.
Brain Research, 661, 163-173. Lang, P. J., Greenwald, M. K, Bradley, M. M., & Hamm, A. O. (1993).
Frijda, N. H. (1986). The emotions. New York: Cambridge. Looking at pictures: Affective, facial, visceral, and behavioral reactions.
Gove, P. B. (1986). Webster's third new international dictionary. Spring- Psychophysiology, 30, 261-273.
field, MA: Merriam-Webster. LeDoux, J. E. (1990). Information flow from sensation to emotion plas-
ticity in the neural computation of stimulus value. In M. Gabriel &
Graham, F. K. (1979). Distinguishing among orienting, defense, and
J. Moore (Eds.), Learning and computational neuroscience: Foun-
startle reflexes. In H. D. Kimmel, E. H. van Olst, & J. F. Orlebeke
dations of adaptive networks (pp. 3-52). Cambridge, MA: Bradford
(Eds.), The orienting reflex in humans. An international conference Books/MIT Press.
sponsored by the Scientific Affairs Division of the North Atlantic Treaty Lee, Y., Lopez, D., Meloni, E., & Davis, M. (1994). A primary acoustic
Organization (pp. 137-167). Hillsdale, NJ: Erlbaum. startle circuit: Role of auditory root neurons and the nucleus reticuloris
Graham, F. K., & Cliffton, R. K. (1966) Heart rate change as a component poutis caudalis. Poster session presented at the Annual Meeting of
of the orienting response. Psychophysiology, 65, 305-320. the Society for Neuroscience, Miami, Florida.
Gray, J. A. (1985). Issues of neuropsychology of anxiety. In A. H. Tuma Mandler, G., Mandler, J. M., Kremen, I., & Sholiton, R. (1961). The
& J. D. Maser (Eds.), Anxiety and the anxiety disorders (pp. 5-26). response to threat: Relations among verbal and physiological indices.
Hillsdale, NJ: Erlbaum. Psychological Monographs, 75(1, Whole No. 513).
Greenwald, M. K., Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1990). Masterson, F. A., & Crawford, M. (1982). The defense motivation system:
The acoustic startle reflex indexes aversive learning [Abstract], Psy- A theory of avoidance behavior. Behavioral and Brain Sciences, 5,
chophysiology, 27, S39. 661-696.
Greenwald, M. K., Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1991). Mehrabian, A., & Russell, J. A. (1974). An approach to environmental
Fear and loathing: Startle reflex and visceral response [Abstract]. Psy- psychology. Cambridge, MA: MIT Press.
chophysiology, 28, S26. Miller, N. (1944). Experimental studies of conflict. In J. M. Hunt (Ed.),
Greenwald, M. K., Cook, E. W., & Lang, P. J. (1989). Affective judgment Personality and the behavior disorders (pp. 431-465). New York:
and psychophysiological response: Dimensional covariation in the Ronald Press.
evaluation of pictorial stimuli. Journal of Psychophysiology, 3, 5 1 - Morris, M., Bradley, M., Bowers, D., Lang, P., & Heilman, K. (1991).
64. Valence-specific hypoarousal following right temporal lobectomy
Hamm, A. O., Cuthbert, B. N., Globisch, J., & Vaitl, D. (in press). [Abstract]. Journal of Clinical and Experimental Neuropsychology,
Startle reflex modulation and psychophysiological response patterns 13, 42.
to affective stimulation in subjects with simple phobias and nonphobic Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure
controls. Journal of Abnormal Psychology. of emotions. Cambridge, MA: Cambridge Press.

384 May 1995 • American Psychologist

Osgood, C , Suci, G., & Tannenbaum, P. (1957). The measurement of Solomon, R. L., & Corbit, J. D. (1974). An opponent-process theory of
meaning. Urbana, IL: University of Illinois. motivation: I. Temporal dynamics of affect. Psychological Review, 78,
Patrick, C. J., Berthot, B., & Moore, J. D. (1993). Effects of diazepam 3-43.
on startle reflex potentiation in human subjects [Abstract]. Psycho- Spence, E. L., Fischler, I., & Lang, P. J. (1995). Encoding and retrieval
phvsiologv, 30, S49. of affective text. Unpublished manuscript.
Patrick, C. J., Bradley, M. M., & Lang, P. J. (1993). Emotion in the Spence, K. W., & Runquist, W. N. (1958). Temporal effects of conditioned
criminal psychopath: Startle reflex modification. Journal of Abnormal fear on the eyelid reflex. Journal of Experimental Psychology, 55,
Psychology, 102, 82-92. 613-616.
Rafferty, J., & Norling, R. (1989). Cricket Graph user's guide. Malvern, Stritzke, W. G. K., Patrick, C. J.,& Lang, A. R. (1995). Alcohol and
PA: Cricket Software. human emotion: A multidimensional analysis incorporating startle
Roldan, E., Alvarez-Pelaez, R., & Fernandez de Molina, A. (1974). Elec- probe methodology. Journal of Abnormal Psychology, 104, 114-122.
trographic study of the amygdaloid defense response. Physiology and Swerdlow, N. R., Caine, S. B., Braff, D. L., & Geyer, M. A. (1992). The
Behavior, 13, 779-787. neural substrates of sensorimotor gating of the startle reflex: A review
Ross, L. E. (1961). Conditioned fear as a function of CS-UCS and probe of recent findings and their implications. Journal of Psychopharma-
stimulus intervals. Journal of Experimental Psychology, 61, 265-273. cology, 6, 176-190.
Russell, J. (1980). A circumplex model of affect. Journal of Personality Tellegen, A. (1985).'Structures of mood and personality and their rele-
and Social Psychology, 39, 1161-1178. vance to assessing anxiety, with an emphasis on self-report. In A. H.
Schlosberg, H. (1952). The description of facial expression in terms of Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders (pp.
two dimensions. Journal of Experimental Psychology, 44, 229-237. 681-706). Hillsdale, NJ: Erlbaum.
Schneirla, T. (1959). An evolutionary and developmental theory of bi- Ursin, H. (1965). Effect of amygdaloid lesions on avoidance behavior
phasic processes underlying approach and withdrawal. In M. Jones and visual discrimination in cats. Experimental Neurology, 11, 298-
(Ed.), Nebraska Symposium on Motivation. Lincoln: University of 317.
Nebraska Press. Ursin, H., Jellestad, F., & Cabrera, I. G. (1981). The amygdala, explo-
Schupp, H., Cuthbert, B. N., Bradley, M. M., Lang, P. J., & Birbaumer, ration and fear. In Y. Ben-Ari (Ed.), The amygdaloid complex (pp.
N. (1993). Imagining the percept: Cortical and peripheral measures 317-329). Amsterdam: Elsevier/North-Holland.
[Abstract]. Psychophysiology, 30, S58. Vrana, S. R., & Lang, P. J. (1990). Fear imagery and the startle probe
Seligman, M. E. P. (1975). Helplessness: On depression, development reflex. Journal of Abnormal Psychology, 99, 189-197.
and death. San Francisco: Freeman. Vrana, S. R., Spence, E. L., & Lang, P. J. (1988). The startle probe
Shaver, P., Schwartz, J., Kirson, D., & O'Connor, C. (1987). Emotion response: A new measure of'emotion? Journal of Abnormal Psychology,
knowledge: Further exploration of a prototype approach. Journal of 97,487-491.
Personality and Social Psychology, 52, 1061-1086. Weiss, J. M. (1989). Electrophysiology of the locus coeruleus: Implications
Simons, R. F., & Zelson, M. F. (1985). Engaging visual stimuli and reflex for stress-induced depression. In G. F. Koob, C. L. Ehlers, D. J. Kupfer
blink modification. Psychophysiology, 22, 44-49. (Eds.), Animal models of depression (pp. 111-134). Boston: Birkhauser.
Sokolov, Y. N. (1963). Perception and the conditioned reflex (S. W. Way- Wundt, W. (1896). Gundriss der Psychologie [Outlines of psychology].
denfeld, Trans.). New York: Macmillan. (Original work published 1958) Leipzig, Germany: Entgelmann.

May 1995 • American Psychologist 385