QEEG Correlates of Film Presentations
Cinematic narratives are by design very complex visual
and auditory stimuli. Many researchers would like to determine
when an individual is paying attention to a film or television
commercial and how deeply the information being processed.
Unfortunately, reliable quantitative measurements of covert
processes such as visual and auditory attention are difficult to
acquire. Evaluation of how well print or film media convey
information have found conflicting results (Cowen, 1984; Furnham,
Proctor, & Gunter, 1988; Siegel, 1973). Audio-visual
presentations usually produce greater encoding of messages
compared to audio presentations alone (Hayes, Kelly, & Mandel,
1986; Siegel, 1973), but sometimes both presentations work
equally as well (Helmreich, 1976). Physiological measures such as
eye heart rate (Lang, 1990) and pupil dilation (Beatty, 1982)
capture dynamics of nonspecific arousal associated with visual
and auditory attention, but evaluation of specific attentional
responses require topographic measures such as quantitative EEG.
Experiment 1: EEG correlates of media involvement
Krugman (1971) was the first to use EEG in media
research. He found that watching television and reading a
magazine activated occipital cortex for similar durations in a
single subject. Walker (1980) later confirmed this finding in 18
subjects. Appel, Weinstein, and Weinstein (1979) also examined
EEG correlates of television viewing and reported parietal
activation when subjects attended to television commercials.
Extensive media research using topographic EEG has been
undertaken by Rothschild and Reeves. Reeves et al. (1985)
demonstrated that alpha power decreased temporarily in response
to character movements and camera edits (i.e., discontinuous
segments) in a television show. Rothschild, Hyun, Reeves,
Thorson, and Goldstein (1988) reported asymmetric occipital
activation to verbal and nonverbal stimuli in television
commercials, in expected directions. Emotional content of
television scenes also engaged specific cortical areas (Reeves,
Lang, Thorson, & Rothschild, 1989).
EEG correlates of film presentations
Of the three major mediums, print, radio, and cinema,
cinematic presentations should be the most engaging. Unlike print
or radio, film and television programs involve two modalities
that must be integrated simultaneously. In addition to literary
conventions of narrative, such as action and dialogue, background
music and visual montage must also be incorporated into a
meaningful sequence. Multiple cortical function will be activated
by films. Parietal, posterior temporal, and occipital cortex are
all activated during story integration (Pockberger et al.,
1985), Right hemisphere mechanisms are said to integrate content
across a narrative (Wapner, Hamby, & Gardner, 1981; Lechevalier,
Petit, Eustache, Lambert, Chapon, & Viader, 1989; Moore, 1987;
Brownell, Michel, Powelson, & Gardner, 1983; Joanette, Goulet,
Ska, & Nespoulous, 1986). Left temporal cortex is involved in
theme abstraction as well as general retention and retrieval of
story and pictorial content (Prevey, Delaney, & Mattson, 1988;
Frisk & Milner, 1990; Zaidel & Rausch, 1981; Pigott & Milner,
1993).
Film presentations also engage perceptual processes that
have EEG correlates. Structural features of the cinematic media
such as camera zooms and movements in video presentations produce
occipital activation (Rothschild & Thorson, 1983; Lang, 1990;
Reeves et al., 1985). Perception and retention of complex
visual information requires engagement of posterior and right
anterior temporal functions in particular and the right
hemisphere in general (Villa, Gainotti, De Bonis, & Marra, 1990;
Gundel & Wilson, 1992; Pigott & Milner, 1993; Makino, 1986). An
attentional focus on environmental stimulation has also been
found to activate temporal and parietal cortex (Cole & Ray,
1985).
In the present study, subjects watched a group of short
films ("previews of coming attractions") and rated subjective
interest for each film using a Likert-type scale (FILM
condition). Approximately one-half of these films were watched a
second time during which subjects rated subjective interest
continuously by manipulating a joystick (RATE condition). It is
predicted that both viewing conditions will engage temporal and
parietal areas compared to the eyes open baseline (EO) condition
due to the external attentional focus and processing of complex
visual and auditory stimuli in films. Integration of cinematic
material into a coherent structure will also engage the right
hemisphere preferentially.
Activation due to joystick movements
Although both viewing conditions require monitoring of
subjective interest, each is achieved through very different
means. One method uses a continuous manual task that involves
self-paced hand and wrist movements and the other consists of
writing a numeric rating on paper after each film's completion
that may be performed retrospectively. Because voluntary
movements of hands or thumbs activate bilateral or contralateral
central areas (Pfurtscheller & Berghold, 1989; Tarkka, & Hallett,
1990; Sterman et al., 1994), any condition with joystick
movements (MC, RATE) is predicted to engage central cortex more
than conditions without manual components (EO, FILM).
Effect of order on viewing films
Each film in the RATE condition was presented
approximately an hour before in the FILM condition. Information
encoded during the first viewing may be retrieved during the
second to facilitate narrative integration. The process of
recollection may diminish attention to the stimulus itself and
result in parietal and temporal lobe disengagement (Ray & Cole,
1985), but the act of retrieval may also engage temporal lobe
mechanisms (Brown, Wilson, & Riches, 1987; Samson & Zatorre,
1992). Accordingly, it is predicted that RATE will be associated
with parietal disengagement compared to FILM.
Summary of predictions
Five conditions of increasing task demand will be
analyzed: two baseline conditions (EC, EO), a motor control task
(MC), and two viewing tasks, one in which subjective interest in
film previews are rated on a scale (FILM) and the other in which
interest is rated continuously during the film presentation
(RATE). The EC condition is predicted to show nonspecific
disengagement compared to EO (see Chapter 3) and all other
conditions in which eyes are open. FILM and RATE will be engaged
at posterior and right-sided cortical areas compared to EO and
MC. MC and RATE will be activated at central cortex compared to
EO and FILM. Finally, RATE will be disengaged at parietal cortex
compared to FILM.
METHOD
Subjects
EEG was acquired from 20 healthy right-handed subjects
(10 male and 10 female) described in Chapter 3.
Film-Viewing Survey
Before subjects watched films, they responded to a film-
viewing survey prior to the run of the study. Subjects were
instructed to circle titles of films he or she had watched in the
past from a list of 27 feature films (see Appendix C).
Cinematic Narratives
Twenty-seven cinematic narratives (previews of coming
attractions), recorded from broadcast TV, were used in this
study. Each narrative lasted 2 min +/- 15 s, was in color, and
consisted of dialogue, narration, and cinematic footage from a
feature film. Subjects watched approximately 15 cinematic
narratives that advertised films which he or she had not seen
previously.
Subjects rated subjective interest in 2-min cinematic
narratives using a Likert-type scale with a range from (1) not
at all interested, and (10) very interested. A 10-
point scale was used due to its familiarity in this context (cf.
Neisser, 1976). Subjects also rated their enjoyment of each film
on a similar scale.
Content Questions
Four types of multiple-choice questions were constructed
to judge relative attentiveness and success in narrative
comprehension (see example, Appendix D). Subjects circled
responses in a response packet consisting of 14 pages, two
narratives per page. Each page contained four multiple-choice
questions per narrative and subjects only responded to those
questions pertaining to those films they had watched.
Continuous Rating of Subjective Interest
A single-axis joystick was used to rate subjective
interest continuously. The far-forward position indicated very
interested in the film and the far-backward position
indicated not at all interested. Absolute joystick
orientation was counterbalanced across subjects. For instance,
pulling towards one's body activated the far-forward position for
some subjects and the far-backward position for other subjects.
Joystick position was sampled once per s and was manually
synchronized with the EEG data. The sampling rate provides more
than enough temporal resolution in that behavioral interesting
ratings changed every 7.55 s on average (see Chapter 6).
EEG methodology
EEG acquisition, preparation, and data analysis were
identical to the method presented in Chapter 3.
Procedure
EC and EO baseline conditions were recorded prior to
viewing any narratives, after the first set of 15 narratives, and
at the end of the study. Subjects rated subjective interest and
enjoyment after watching each narrative on two Likert-like
scales. After the first set of narratives, subjects answered
content questions to verify attentiveness during film
presentations. After completing this test, subjects manipulated a
joystick at their own pace for 1 min with either the left or
right hand (MC). Subjects then watched 3 or 4 films during which
they manually rated subjective interest in the film. Following
this, subjects again manipulated the joystick at their own pace
with the other hand for 1 min and then watched and continuously
rated an additional 3 or 4 more films. Hand order was
counterbalanced across subjects.
RESULTS
Two narrative integration tasks (FILM and RATE) were
compared to three baseline or control conditions (EC, EO, and MC)
in four spectral parameters. Task means of approximately 60
epochs (2 min) were compared using univariate ANOVAs. As in
Chapter 3, data were initially pooled into six functional regions
and degrees of freedom were adjusted to compensate for
nonsphericity (Huynh & Feldt, 1976; Vasey & Thayer, 1987).
Although no correction was made on the probability level for
independent multiple ANOVAs, any finding in LMAGN, SD, or RV, can
be corroborated in another spectral parameter to be judged
reliable.
Task demand in tasks and topography
Condition means are presented in Table 4.1. A main
effect of tasks was seen in each spectral parameter (see Table
4.2.1). EC differed from all conditions in all but one
comparisons to such an extent that further analyses concentrated
on tasks in which subject's eyes were open.
Table 4.1. Mean spectral values of five conditions (19 sites, n=20)
LMAGN SD SLOPE RV
--------------------------------------------------------
EC 2.00 3.38 -0.012 0.023
EO 1.41 1.92 0.010 0.013
MC 1.41 1.81 0.007 0.045
FILM 1.35 1.62 0.002 0.013
RATE 1.33 1.52 0.005 0.013
--------------------------------------------------------
Tasks can be divided into deactivated (EO, MC) and
activated (FILM, RATE) conditions. LMAGN differed significantly
between EO and both viewing conditions and between MC and RATE
[p<.05]. Standard deviation declined
Table 4.2. Statistical comparisons of five conditions
1. Main effect 2. Interaction
--------------------------------------------------------
FLMAGN (4,72)= 84.366, p<.05 FLMAGN(3,58)=23.475, p<.05
FSD (4,72)= 76.180, p<.05 FSD (3,53)=31.576, p<.05
FSLOPE (4,72)= 2.820, p<.05 FSLOPE(2,37)= 3.126, p=.056
FRV (4,72)= 70.093, p<.05 FRV (2,34)= 8.905, p<.05
--------------------------------------------------------
consistently with greater task demand [p<.05] (see Figure
4.1); only EO and MC did not differ [FSD(1,18)= 0.812,
ns]. Mean SLOPE values ranged from 0.002 æV (FILM) to 0.010 æV
(EO) per epoch across conditions, but no comparison reached
significance [p>.05]. RV was similar between all
conditions except for MC, which was greater than all other
conditions [p<.01].
Figure 4.1. Variability associated with four conditions
(n=20, 19 sites, * p<.05).
Functional activation by tasks
Task by recording site interactions were observed in all
parameters (see Table 4.2.2). Frontal pole values were especially
vulnerable to eye blink artifact and were ignored in later
analyses. Significant differences at fronto-temporal sites for MC
were also ignored on similar grounds.
Eyes open baseline and motor control conditions
Task by recording site interactions were found between
EO and MC [FLMAGN(2,44)= 6.947, p<.01;
FRV(1,28)= 6.174, p<.01]. Figure 4.2 illustrates
how MC resulted in lower LMAGN at site C4 (p<.01) and Cz
(p<.03).
Figure 4.2. Central activation associated with joystick
movement (n=20, * p<.01 + p<.05).
Control and film conditions
A task by recording site interaction was seen between EO
and FILM in one parameter only [FSD(1,19)= 4.816,
p<.05]. FILM was less variable at site Fz and all
posterior sites [p<.01]. A task by recording site
interaction was also seen between MC and FILM [FLMAGN(2,43)=
6.241, p<.05; FSD(2,41)= 4.631, p<.05;
FRV(1,27)= 7.898, p<.05] and between MC and RATE
[FLMAGN(2,41)= 3.218, p<.05; FSD(2,40)= 3.488,
p<.05; FRV(1,25)= 7.014, p<.05]. MC was
deactivated at frontal cortex compared to RATE [p<.01]. MC
was more variable at site O2 than either film condition
[p<.01] and at site F3 for RATE [p<.01].
Significant effects of film presentation between narrative and
control conditions are shown in Figure 4.3.
Rating and viewing film tasks
A viewing condition by recording site interaction
approached significance in amplitude [FLMAGN(2,42)= 3.088,
p=.06; FSD(2,31)= 1.070, ns; FSLOPE(2,36)=
0.920, ns; FRV(2,32)= 1.047, ns]. RATE was activated at
sites T3 and T4 compared to FILM [p<.01].
Laterality patterns across conditions
EC values were included in an analysis of functional
asymmetries. A condition by lateral site difference interaction
approached significance in LMAGN [FLMAGN(4,72)= 2.407,
p=.056; FSD(4,70)= 1.584, ns; FSLOPE(3,63)=
1.534, ns; FRV(2,35)= 1.487, ns].
Figure 4.3. Topographic activation associated with film
presentations (n=20).
As shown in Figure 4.4, EC resulted
in left posterior temporal activation (site T5) and FILM resulted
in analogous right activation (site T6) [F(1,19)= 8.705,
p<.01]. Subjects were bilaterally activated in these areas
for EO and did not differ from either condition [p>.02].
Figure 4.4. Functional asymmetry associated with baseline
and film presentation (n=20, * p<.01).
DISCUSSION
Film presentations resulted in nonspecific arousal as
well as selective attention. Except for anterior temporal
activation, viewing films a second time resembled first
presentations in all spectral parameters.
Engagement in response to motoric movements
The motor control task consisted of self-paced movements
of the hand and wrist. As expected, central cortex was engaged
during this condition compared to an eyes open baseline. Alpha
variability, as measured by RV, was higher at all sites for the
motor control condition, indicating decreased arousal more than
functional activation. Slope values were highly variable between
subjects, especially for the motor control task. Neither the
motor control nor rating task differed from the FILM condition at
central cortex, in contrast to predictions.
EEG correlates of film presentations
Films consist of complex visual and auditory patterns
which changed rapidly and often in unexpected ways to keep an
audience's interest. Main effects between film and control
conditions reflect how a general state of arousal or
attentiveness was essential for perception and integration of the
abundant information contained in each film.
Most story elements transpire within a temporal context
and are interpreted with reference to a time line. Maintaining a
chronology of events is required for interpreting plot and
characterization. Watching a film in either condition (FILM,
RATE) resulted in frontal activation. Frontal areas are involved
in monitoring temporal order of stimuli (Milner & Petrides, 1984;
Villa et al., 1990) as well as sustaining attention in a
task (Rezai, Andreasen, Alliger, Cohen, Swayze, & O'Leary, 1993;
Wilkins, Shallice, & McCarthy, 1987). SD and RV values paralleled
the LMAGN finding.
Functional differences between viewing conditions
The continuous rating task differs from the film viewing
condition in three ways. It includes a motor component, a
continuous monitoring component, and a memory component beyond
normal processes elicited by film presentations. Bilateral
activation of anterior temporal cortex was seen during the
continuous rating task compared to the viewing condition.
Activation of temporal cortex is unlikely to be a result of
manipulating a joystick in that simple motoric behaviors are
almost exclusively associated with central cortical activation
(Kutas & Donchin, 1980; Pfurtscheller, 1992).
For the rating task, a subject was required to monitor
subjective interest constantly. A subject in the film condition
judged interest in films in a familiar manner and pace: subjects
provided numerical ratings after film completion and paid
attention to the material however they thought fit. Subjects in
the rating task had to attend to every scene to some extent in
order to determine whether subjective interest was increasing,
decreasing, or remaining stable with each scene. The sheer
frequency of interest judgments required may have compelled
subjects to develop novel strategies of processing film content.
An easy and effective strategy, described by many subjects after
task completion, was to compare each scene to those preceding it.
Interest level was judged locally: is this scene more or less
interesting than the one before it and by how much? This strategy
essentially adds a short-term memory component to interest
monitoring, which may explain the anterior temporal activation.
Temporal lobes and adjacent limbic areas are crucial for short
term memory (Brown, Wilson, & Riches, 1987; Eskandar, Richmond, &
Optican, 1992).
Given the complexity of cinematic materials, it is
likely that subjects relied on recollection of visual and
auditory information from the initial viewing to integrate the
second viewing. Such a strategy might not only reduce task demand
but enhance story enjoyment. Whether subjects engaged memory
processes to perform the continuous monitoring task, or to assist
narrative integration, or both, is not clear. Both temporal lobes
are critical for retrieving visual and auditory information
(Tucker, Novelly, Isaac, & Spencer, 1986; Samson & Zatorre, 1992)
and short term memory may be involved in either component of the
rating task.
It is interesting that the differences in temporal
activation was bilateral. One would expect activation in right
temporal lobes during film presentations. The right temporal lobe
plays an important role in narrative integration and processing
visual information (Lechevalier et al., 1989; Wapner et
al., 1981). The right temporal lobe, particularly the right
hippocampus, is implicated in visual imagery tasks (Jones-Gotman
& Milner, 1978) and in preserving figurative detail and spatial
composition of complex visual scenes (Pigott & Milner, 1993).
Films also contain verbal information, which may explain the
additional left temporal activation. Elements of verbal content
are processed in left anterior temporal cortex (Samson & Zatorre,
1992).
Lateralization during control and task conditions
Laterality differences approached significance for
posterior temporal areas. Resting with eyes closed resulted in
left posterior temporal activation, viewing films in right
posterior temporal activation, and eyes open resting showed
bilateral activation. Moss, Davidson, and Saron (1985) reported
left anterior temporal and parietal activation in eyes closed
conditions and bilateral activation during eyes open conditions
for Westerners. Right activation during film viewing may also
reflect processing of complex visuospatial patterns (Bradshaw &
Nettleton, 1981) and/or integration of story elements into a
meaningful whole (Wapner et al., 1981).
Experiment 2: Gender effects in topographic EEG
Most psychophysiological research of gender effects
concentrates on functional laterality. McGlone (1980) reviewed
studies that examined gender as a factor and concluded that males
exhibited greater functional asymmetry than females for most
tasks. But clinical findings are inconclusive. Although Kimura
(1983) reported that males were more often stricken with aphasia
after left-sided lesions than females, recent reports have found
no evidence of a gender bias in aphasia incidence (Kertesz &
Thomas, 1989; Scarpa, Colombo, Sorgato, & de Renzi, 1987).
Crossed aphasia is reported in similar numbers of males and
females (McGlone, 1980), suggesting that right hemisphere
dominance for speech functions is not, as expected, more common
in women. Some researchers find stricter functional segregation
for males (Lake & Bryden, 1976; McGlone, 1978; Sundet, 1986;
Inglis & Lawson, 1982) or more bilateral representation of
functions for females (Turkheimer & Farace, 1992), but others
find no gender differences (Herring & Reitan, 1992; Herring &
Reitan, 1986; Scarpa et al., 1987).
Although neuroanatomical differences exist between men
and women (Aboitiz, Scheibel, Fisher, & Zaidel, 1992; Clarke,
1990), gender effects, when present, may not necessarily imply
physiological or anatomical differences per se. Gender
effects may signify differences in development (Shearer, Cohn,
Dustman, & LaMarche, 1984; Brown & Grober, 1983), task
characteristics (Earle & Pikus, 1982; Shepherd, 1982), or
cognitive strategies (Kinsbourne, 1980; Faber-Clark & Moore,
1983; Inglis & Lawson, 1982; Sundet, 1986; Zaidel, Aboitiz,
Clarke, Kaiser, & Matteson, in press). Berfield, Ray, and
Newcombe (1986) argued that differences arise not from the
biological sex of a participant but from differences in sex role
orientation (i.e., masculinity, femininity).
A survey of recent EEG research does not clarify the issue.
Although many EEG studies found asymmetrical activation in males
and/or bilateral activation in females (Rippon, 1990; Trotman &
Hammond, 1979; Tucker, 1976; Beaumont, Mayes, & Rugg, 1978; Flor-
Henry & Koles, 1982), other researchers described diametrical
(Moore & Haynes, 1980; Rebert & Mahoney, 1978) or negligible
differences (Walker, 1980; Shepherd, 1982). Other EEG researchers
have demonstrated main effects of gender, such as greater alpha
power and higher peak frequencies for females, without finding
correlates of functional asymmetry (Etevenon, 1986; Shepherd,
1982; Deakin & Exley, 1979).
Whether gender effects reflect actual physiological and
anatomical differences between genders or task-related
differences is unresolved. Systematic EEG investigation of gender
effects has been performed by a handful of researchers and
further efforts are required before EEG research and its
paradigms will be able to weigh in decisively on this issue. One
conclusion that can be drawn from this body of research is the
need for stringent controls, especially for task characteristics
and subject strategies. Nevertheless, for this experiment it is
predicted that males will be more engaged in the right hemisphere
and females will be more bilaterally engaged during film
presentations (e.g., Rippon, 1990; Trotman & Hammond, 1979).
METHOD
Date were recalculated for each gender (10 male, 10
female).
RESULTS
Gender effects in EEG were investigated across five
conditions. EC, EO, and MC conditions were collapsed across
replications and FILM and RATE conditions were averaged across
numerous films, controlling for subjective interest. Spectral
parameters of alpha activity (8-12 Hz) were analyzed using
univariate ANOVAs. Data were pooled into six functional areas and
degrees of freedom were adjusted.
Gender effects in tasks and topography
No main effects or interactions of cortical site were
observed for any spectral parameter [p>.05]. A marginal
gender effect in functional asymmetry was found for EO in
amplitude [FLMAGN(4,68)= 3.215, p<.05]. Males were
relatively more activated in right fronto-temporal areas (F7-F8)
than females, but it did not reach significance
[FLMAGN(1,18)= 5.374, p=.03]. Significant differences
in laterality were found for FILM, and in additional parameters
[FLMAGN(4,78)= 3.762, p<.05; FSD(4,65)= 2.737,
p<.05; FSLOPE(4,66)= 0.539, ns; FRV(4,69)=
2.514, p<.05]. As shown in Figure 4.5, when subjects
viewed films, males were more activated in right anterior cortex
and females were more activated in left anterior cortex [F7-F8: FLMAGN(1,18)= 7.703, p=.01;
F3-F4: FLMAGN(1,18)= 6.120, p=.02; T3-T4:
FLMAGN(1,18)= 6.353, p=.02]. Fronto-temporal
differences in amplitude were corroborated in variability and
trend instability [F7-F8: FSD(1,18)= 7.801, p=.01;
FRV(1,18)= 7.264, p<.02].
Figure 4.5. Gender differences in anterior functional
asymmetry during film presentation (n=10 male, 10 female, *
p=.01, + p=.02).
DISCUSSION
As expected, genders did not differ in arousal or
regional engagement patterns for baseline or viewing conditions.
Few researchers have obtained main effects or topographic
interactions with gender in alpha activity (exceptions include
Etevenon, 1986; Shepherd, 1982; Deakin & Exley, 1979).
Gender differences in laterality
Although regional patterns did not differ between males
and females in any condition, laterality differences were found
in anterior cortical areas. Figure 4.6 illustrates frontal
engagement across a 2-min film for males and females. During film
presentations males were activated in right fronto-temporal
cortex and females were activated in left anterior temporal
cortex. Flor-Henry and Koles (1982) reported similar asymmetries
during baseline conditions. Right frontal cortex is especially
critical for sustaining attention (Wilkins et al., 1987)
and is involved in processing nonverbal/visual information such
as abstract and representational drawings (Petrides & Milner,
1982; Milner & Petrides, 1984). Left temporal cortex is involved
in verbal processing (Samson & Zatorre, 1992). These differences
may indicate cognitive strategies employed in interpreting films
(Faber-Clark & Moore, 1983).
Figure 4.6. Example of frontal engagement specific to
each gender during film viewing. Note the modulating, deactivated
site is F8 for females and F7 for males.
The above finding suggests that women concentrate more
on verbal information in each film and men focus more on the visual
content. Alternately, these differences could emerge at later
processing stages: women may be more analytical and men more
holistic in how they interpret cinematic material. Analysis of
content questions determined that men and women did not differ in
identifying important verbal and visual stimuli contained in
these films; thus indicating the gender differences do not
reflect perceptual strategies (i.e., focusing on one type of
information or modality over another), but rather develop as a
result of post-perceptual interpretative strategies.
Experiment 3: EEG correlates of content recall
Topographic EEG may be used to evaluate depth of
processing of a film or television program. Reeves et al.
(1985) reported relationships between alpha activity and recall
and recognition accuracy of commercial content. Alpha amplitude
at occipital cortex correlated with how well each subject was
able to recall or recognize product (general) information,
specific scenes, and visual detail (package characteristics).
Rothschild and Hyun (1990) determined that alpha attenuation
persisted at occipital cortex (only region investigated) after
presentation of high recognition elements. Also, greater right
hemispheric activation occurred during the onset of high
recognition stimuli, followed by a left hemisphere activation
during post-perceptual processing.
Subjects responded to four questions about each film
they watched. It is predicted that accurate identification of
general and specific content will activate posterior cortex and
identification of visual items will result in right hemispheric
activation followed by left hemispheric activation. The fourth
questioned determines whether subjects could judge the relative
length of each film.
METHOD
Subjects, materials, and procedure are described above.
Four types of questions were asked for each film watched by the
subject. The first question (Q1) concerned general content of a
film, such as who was the titled character or what was the film
about? The second question (Q2) required subjects to recognize an
important association with theme, plot, or characterization, such
as the emergence of scandal or character defiance. The third
question (Q3) required subjects to identify which one of six
objects or persons appeared in the film. Targeted items were
peripheral to plot and characterization and were present on the
screen for less than 4 s. The fourth and final question (Q4)
required subjects to judge the relative length of each film using
three comparative categories: shorter, longer, and average
duration (scored as +/- 1 standard deviation from the mean).
RESULTS
Each question was analyzed separately. Spectral
parameters of alpha activity (8-12 Hz) were derived across all
epochs of each film. As before, degrees of freedom were adjusted
to compensate for nonsphericity.
Subjective interest and content recall
Response accuracy was calculated for high, medium, and
low interest narratives for all questions. General content was
best recognized in high interest films (96.3%) but worst in
medium interest films (65.0%), with low interest films (84.2%)
falling in between [F(2,34)= 20.955, p<.05]. A similar
pattern was seen for specific questions [F(2,37)= 3.857,
p<.05]. No effect of interest was found for object
identification nor time estimation [p>.05].
EEG correlates of content recognition
All significant differences were found in SLOPE only.
Main effects were observed for the general question (Q1)
[FSLOPE(1,19)= 4.304, p=.05]. Subjects were less
accurate to Q1 when they exhibited considerable positive trends
during the film. A main effect and interaction of laterality was
also found for Q1 [FSLOPE(1,19)= 9.913, p=.05;
FSLOPE(3,53)= 3.451, p<.05]. General knowledge of a
film was associated with smaller positive slopes in the right
hemisphere, particularly at T5 [p<.01] and P3
[p<.02].
A recording site interaction was found for the specific
question (Q2) [FSLOPE(4,83)= 2.944, p<.05]. Accurate
identification of critical elements was associated with larger
positive slope coefficients at site Cz [p<.01].
Temporal dynamics of EEG and stimulus presentations
Questions 2 and 3 tested subjects' knowledge about
specific elements which appeared at specific times during each
film. The above analyses involved condition means only, a global
measure that summarized across more than 2 min of data. In the
following analyses, EEG amplitude immediately before, during, and
after stimulus presentations were analyzed. A subset of the
subject group (6 for Q2, 2 for Q3) were excluded from analyses as
they responded accurately to all relevant questions (100%).
Analysis focused on epoch magnitude of temporal and
parietal cortex. Subjects were more accurate when they
deactivated immediately after important scenes were
presented [Fafter(1,13)= 5.948, p<.05]. This effect
was strongest at site P3 [Fafter(1,13)= 10.413,
p<.01]. Subjects were also more accurate when they
activated during presentations of important elements then
immediately deactivated once the scene had passed [Fduring-
after(1,13)= 8.876, p<.05]. As shown in Figure 4.7, this
effect was prominent at sites P3 and Pz [p<.01].
Figure 4.7. Mean magnitude during (0 to 4 s) and
immediately after (4 to 8 s) presentation of important scenes (*
p<.01).
A main effect of accuracy approached significance for Q3
during presentations [Fduring(1,17)= 4.118, p=.058].
Subjects were more accurate in identifying peripheral items if
they were more activated during the actual stimulus presentation.
DISCUSSION
Interest itself was not predictive of content question
accuracy. Significant differences in general and specific
questions were found, but lacked face validity in that subjects
performed most poorly on questions about medium interest
films. Increased arousal, especially of the right hemisphere, was
associated with accurate responses to general content questions
as were trend values for left temporal and, to a lesser extent,
left parietal areas.
General content integration
Subjects had to be moderately aroused to extract the
gist or general thrust of a cinematic narrative. The nonspecific
drop in arousal, associated with the poor performance, may also
reflect internal processing at the expense of integrating
environmental stimuli (Ray & Cole, 1985). Greater attention to
visual information may be required to perceive the gist of a film
(Rebert & Low, 1978; Dujardin, Derambure, Defebvre, Bourriez,
Jacquesson, & Guieu, 1993; Boiten et al., 1992).
Preferential engagement of right parietal and posterior temporal
cortex is associated with visual memory (Makino, 1986) and
interpreting narrative as well as configural information (Rehak,
Kaplan, Weylman, Kelly, Brownell, & Gardner, 1992; Bogen & Bogen,
1983; Lechevalier et al., 1989).
Important event identification
Progressive disengagement at the mid-central cortex
coincided with accurate identification of important scenes in
each film. This finding is unexpected and may reflect reduced
fidgeting and body position changes as a subject becomes more and
more emotional and intellectual absorbed in a film. Stratton &
Zalanowski (1984) reported that the degree of liking in music was
positively correlated with overall relaxation. Individuals
reported less distractions during relaxed periods as well.
Temporal EEG correlates of interest: ERD and PRS
sequences
Subjects were most accurate when they activated at left
and mid parietal areas during important events, then immediately
deactivated. This illustrates event-related desynchronization
(ERD). When relevant stimuli are presented, initial activation of
specific cortical areas is often observed (Pfurtscheller, 1992),
followed by a "refractory" period in which cortical areas
synchronize in counteraction to a previous cognitive response
(called post-response synchronization or PRS). An ERD may
coincide with an orientation reflex (Barry, 1976), which have
been used to facilitate information acquisition, especially
during low involvement conditions (Zillman, 1982). ERDs at left
parietal cortex indicates greater analytical evaluation of
important narrative events. A PRS of parietal cortex was also
prominent in alpha amplitude 4 s after important event
presentation. This sequence of ERD and PRS are essentially slope
changes in alpha activity and alpha trends may be used to
consider whether material is acquired or not.
Recognition of nonessential visual details was
associated with left hemisphere activation. Brain potentials
(EEG, ERP, ERD) recorded from the left hemisphere correlate with
linguistic performance (Gevins, 1986). Konovalov and Otmakhova
(1983) found marked suppression in left hemisphere with rapid
recovery in the right hemisphere for a verbal task. They found
the opposite pattern for a nonverbal task, but to a weaker
extent. It may be the case that only those peripheral items or
details which were verbally encoded during film viewing were
encoded deep enough to be retrieved an hour later whereas the
majority of detail were only encoded visually and thus forgotten.
Attentional fluctuations can also explain this result.
As noted above, externally-directed tasks such as visual judgment
tasks activate left hemisphere cortex, particularly posterior
areas (Ray & Cole, 1985). Narrative integration does not involve
constant rates of visual and auditory processing, but varies
depending on conscious evaluation of interest, novelty,
complexity, and other factors (Reeves et al., 1985;
Hawkins et al., 1991). Accurate identification of
peripheral items may have coincided with greater perceptual
intake than usual during the 4 s that the item appeared on the
screen. The role of attentional variability in narrative
integration will be examined in Chapter 6.
|
