Available online at www.sciencedirect.com
Brain and Language 103 (2007) 304–307
www.elsevier.com/locate/b&l
Commentary
Hemispheric asymmetries in visual word-form processing:
Progress, conflict, and evaluating theories
Chad J. Marsolek *, Rebecca G. Deason
Department of Psychology, University of Minnesota, 75 East River Road, Minneapolis, MN 55455, USA
Accepted 12 February 2007
Available online 6 April 2007
Abstract
The ubiquitous left-hemisphere advantage in visual word processing can be accounted for in different ways. Competing theories have
been tested recently using cAsE-aLtErNaTiNg words to investigate boundary conditions for the typical effect. We briefly summarize this
research and examine the disagreements and commonalities across the theoretical perspectives. This work may provide a good example
why a multi-level, multi-method, and multi-paradigm approach holds the greatest promise for rapid theoretical progress.
Ó 2007 Elsevier Inc. All rights reserved.
Keywords: Functional hemispheric asymmetries; Visual word recognition; Case alternation; Categories; Exemplars
A ubiquitous finding in neuropsychology is the lefthemisphere (LH) advantage in processing visual word
forms. The question of whether it reflects a LH-lateralized
lexicon or dissociable neural subsystems of visual-form
processing is fundamentally important. If visual word processing is performed in qualitatively different ways in parallel subsystems, rather than in a single lexical system, a
greater understanding of the subsystems and their properties may be crucial for explaining the flexibility and complexity of visual word processing. Different effects may be
observed depending on which subsystem is responsible
for performance.
Deason and Marsolek (2005; hereafter D & M) used the
subsystems theory to predict critical boundary conditions
for the LH advantage. This theory posits that an
abstract-category visual subsystem operates more effectively than a specific-exemplar visual-form subsystem in
the LH, and vice versa in the right hemisphere (RH). An
abstract subsystem recognizes the visual-form category of
a word (e.g., the visual-form category for band/BAND),
*
Corresponding author. Fax: +1 612 626 2079.
E-mail address: [email protected] (C.J. Marsolek).
URL: http://levels.psych.umn.edu (C.J. Marsolek).
0093-934X/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.bandl.2007.02.009
whereas a specific subsystem recognizes the visual-form
exemplar of a word (e.g., the exemplar ‘‘band’’ which is different from ‘‘BAND’’). Both subsystems can recognize
word forms, but the typical LH advantage occurs because
the nature of most word processing tasks—what word is
this?, or is it a word?, not which exemplar is it?—typically
advantages an abstract subsystem in the LH (see Burgund
& Marsolek, 1997). D&M theorized that an important
boundary condition may exist, such that the LH advantage
may be eliminated when words take the shape of unfamiliar
wholes (i.e., their holistic configurations are unlikely to
have been viewed before), as occurs with cAsE-aLtErNaTiNg words. This is because a specific subsystem processes
novel visual wholes effectively (Marsolek, Schacter, &
Nicholas, 1996), helping that subsystem to overcome the
typical advantage held by an abstract subsystem.1 D&M
1
Indeed, it is important to emphasize that evidence for an abstract
subsystem operating effectively in the LH and a specific subsystem
operating effectively in the RH should not be found in all experimental
conditions. According to the subsystems theory, both stimulus and task
demands predictably affect the relative contributions of these subsystems
to performance (for examples see Burgund & Marsolek, 1997; Marsolek,
1999; Marsolek & Burgund, 2003; Marsolek & Hudson, 1999).
C.J. Marsolek, R.G. Deason / Brain and Language 103 (2007) 304–307
reported evidence supporting this prediction in two experiments; the LH advantage in word processing (found for
lowercase and for UPPERCASE words) was eliminated
or numerically reversed when novel visual wholes were
used as stimuli.
Ellis, Ansorge, and Lavidor (this issue; hereafter EAL)
disagreed with the interpretation of D&M’s results, and
they reported new experiments to test the main question.
It is notable that D&M had to present cAsE-aLtErNaTiNg words for long exposure times in order to equate word
identification performance for them compared with lowercase/UPPERCASE words. Thus, EAL examined whether
a LH advantage for lowercase words and its elimination
for cAsE-aLtErNaTiNg words would be observed when
lowercase words are presented for the same, relatively long
exposures used for cAsE-aLtErNaTiNg words (144 ms) in
D&M. In line with D&M’s theory, EAL’s results (Experiment 1B) replicated D&M’s. It is worthwhile to note
points of agreement between EAL and D&M. Not only
was a common pattern of results observed across experiments, but EAL also suggest that their loss of a LH advantage when lowercase words were presented very briefly in
Experiment 1A may be due to the hemispheric differences
in visual processing (involving high versus low visual-spatial frequency information; Christman, 1989; Sergent &
Hellige, 1986) that we have cited for the same reasons
(Marsolek, 1999; Marsolek & Burgund, 1997, 2003; Marsolek & Hudson, 1999). However, it also is important to
consider the disagreements between EAL and D&M to
assess progress in this area.
EAL claimed that an important concern with D&M’s
theory is that the features-based processing hypothesized
to occur in an abstract subsystem should enable it to
process cAsE-aLtErNaTiNg words as effectively as lowercase/UPPERCASE words, and hence should lead to
a LH advantage for cAsE-aLtErNaTiNg words. EAL
write, ‘‘Mixed case words contain the same abstract
visual features that the abstract-category subsystems
use to recognise words in familiar formats. Hence it
could be argued that identifying words in mixed case
should be easier for the abstract-category subsystems
than for the specific-exemplar subsystems. . .’’ This is a
misunderstanding of D&M’s theory. EAL have it right
that cAsE-aLtErNaTiNg words contain the same features that an abstract-category subsystem uses to recognize familiar-format words, but it is incorrect to
assume that the extraction of those features is as easy
for cAsE-aLtErNaTiNg words as it is for familiar-format
words (an assumption that has never been made by proponents of the dissociable subsystems theory). Indeed, we
would cite evidence against this assumption. For example, a large literature on contextual cueing effects (e.g.,
Chun & Jiang, 1998, 1999; Jiang, Song, & Rigas, 2005)
indicates that extracting a target shape within configurations of non-target shapes is incidentally learned to be
faster after the configurations have been familiarized
than before they have been familiarized, a visual learning
305
effect that is not accompanied by explicit memory for the
repeated configurations. Our view is that such evidence
from outside the typical domain of word processing is
crucial for efficient theoretical progress.
The results from EAL’s first experiments are consistent
with D&M’s theory, but the results from EAL’s Experiment 2 are not, thus it is perhaps most important to highlight a serious methodological question raised in the
latter. EAL correctly reviewed that under many conditions the subsystems theory leads to the predictions that
case-specific repetition priming (greater priming from
DEAR to DEAR than from dear to DEAR) should be
observed when test stimuli are presented directly to the
RH and no case-specific priming (no greater priming from
DEAR to DEAR than from dear to DEAR) should be
observed when test stimuli are presented directly to the
LH (for evidence see Burgund & Marsolek, 1997; Marsolek, Kosslyn, & Squire, 1992; Marsolek, Squire, Kosslyn, & Lulenski, 1994; Marsolek, 2004). To test these
predictions in a new way, EAL used a (short-term)
masked priming procedure (following Bowers & Turner,
2005), in which a ‘‘prime’’ word appears briefly (60 ms)
in the same location as a subsequently presented target
word (100 ms). The ‘‘prime’’ and target word could be
the same word in the same case-alternating version (e.g.,
DeAr followed by DeAr) or the same word in cross-case
versions (e.g., DeAr followed by dEaR). No greater identification of target words occurred in the same-case condition than in the cross-case condition when stimuli were
directly to the RH, in contrast with the subsystems prediction. However, a prohibitive concern is that, in the
same-case condition, there are no discernibly different
prime and test stimuli—a single word was presented for
160 ms. This means that the target stimuli were presented
for 160 ms in the same-case condition but for only 100 ms
in the cross-case condition, which is crucial because
EAL’s first experiment indicates that RH processes are
relatively advantaged by brief test presentations and relatively disadvantaged by long test presentations. Thus, the
reason why same-case and cross-case performance in
Experiment 2 was equivalent following RH presentations
may be that the briefer presentations of targets in the
cross-case condition benefited RH processing and the
longer presentations of targets in the same-case condition
hurt RH processing, to points at which no difference was
observed between the two conditions following RH presentations. Perhaps more important, another concern is
that case-specific masked priming has been observed by
one of the EAL authors when primes were presented centrally and targets were presented laterally (Lavidor, 2002),
yet no case-specific priming was observed in EAL’s
Experiment 2 when the stimuli were presented in the same
location. This provides evidence that whether primes and
targets are presented in the same location greatly affects
the patterns of results in masked priming experiments.
Until future research clarifies this (potentially very interesting) inconsistency, it is difficult to know how the
306
C.J. Marsolek, R.G. Deason / Brain and Language 103 (2007) 304–307
masked priming paradigm can be used to address casespecific priming.
EAL’s Experiment 2 notwithstanding, which theoretical
approach should be taken to account for the consistent
results in D&M and EAL? Given that both theories can
account for these results (in different ways), a good move
is to appeal to evidence from additional paradigms and procedures. We applaud EAL for making this kind of move by
explaining how their results can account for word length
effects as well. For lowercase and UPPERCASE words,
word lengths (numbers of letters) affect processing following RH presentations to a greater degree than following
LH presentations (Bub & Lewine, 1988; Ellis, 2004; Ellis,
Young, & Anderson, 1988; Lavidor, Ellis, Shillcock, &
Bland, 2001). However, for cAsE-aLtErNaTiNg words
(and nonwords), comparable word length effects are found
following LH and RH presentations (Bub & Lewine, 1988;
Ellis et al., 1988; Lavidor & Ellis, 2001; Lavidor, Ellis, &
Pansky, 2002; Young & Ellis, 1985), thus EAL hypothesize
that a LH lexicon is engaged only when familiar-format
words are presented directly to the LH (novel format words
and words presented to the RH are processed without the
benefit of top-down lexical processes).
An alternative account for word length effects can be
offered from the subsystems theory (Marsolek, 2004),
although future research is needed to provide empirical
tests. With familiar-format words, greater word length
effects following direct RH presentations than following
direct LH presentations may reflect the difference in
amount of visual information stored in whole-based representations (in a specific subsystem) versus featuresbased representations (in an abstract subsystem). The
whole-based representation of a word form (useful for
recognizing the exemplar) necessarily contains more
information than the features-based representation of
that form (useful for recognizing the category of the
form). The longer the word, the more time-consuming
the match to stored representations, for both specific
and abstract subsystems, but that effect should be greater
in a subsystem that has representations containing relatively more information than in a subsystem that has
representations containing relatively less information
(the more the represented information, the more timeconsuming the match to that information). Note that
this explanation applies to standard words. When nonwords are presented to be identified, they do not match
very well the preexisting representations in both abstract
and specific subsystems, and apparently different processing compared with the processing that normally occurs
in either abstract or specific subsystems (perhaps letterby-letter processing) takes place to cause the length
effects. When cAsE-aLtErNaTiNg words are presented,
both abstract and specific subsystems should have a high
degree of difficulty in matching the input stimuli to
stored representations, given the novelty of the input
forms. The increased difficulty should be exacerbated
by longer words, especially in the abstract subsystem
(in the specific subsystem, the more effective processing
of novel whole forms should counteract the impairment),
causing word length effects in an abstract subsystem to
be comparable with those in a specific subsystem.2
It is not surprising to find that different cognitive theories can account for a common set of behavioral results
in different ways. In fact, Anderson (1978) provides an
excellent discussion of this kind of situation and an argument that behavioral evidence alone may be insufficient
for ultimately arbitrating between the relevant cognitive
theories. Our perspective is like EAL’s in that we value
appeals to evidence from additional paradigms, but we
go further. For explanatory adequacy of a theory of visual
form processing (see Marsolek, 2003), mutual constraints
on theory from multiple levels of analysis, multiple methods, and multiple kinds of visual-form stimuli likely are
needed for greatest theoretical progress. Any one of these
kinds of constraints has its limitations or weaknesses, but
converging evidence from different levels, methods, and
stimuli can help to overcome such limitations and weaknesses. For example, evidence from object processing also
is used to constrain the subsystems theory (e.g., Burgund
& Marsolek, 2000), in part because neuropsychological evidence (Farah, 1991) and neuroimaging evidence (Price &
Devlin, 2003) indicates that common visual processing in
the brain underlies normal visual word and object identification. Also, evidence from interhemispheric transfer
(Marsolek, Nicholas, & Andresen, 2002), neuroimaging
(e.g., Koutstaal et al., 2001), brain-damaged patients (e.g,
Vaidya, Gabrieli, Verfaellie, Fleischman, & Askari, 1998),
and neuromodulatory effects (Burgund, Marsolek, & Luciana, 2003) are used to support the subsystems theory.
These multiple and varied constraints may click together
to derive an additional ‘‘emergent’’ constraint if an integrated ‘‘whole’’ is greater than the sum of its ‘‘parts,’’
essentially helping to explain why the visual system appears
to work in that way and not in other potential ways. This is
the approach we recommend when faced with multiple cognitive theories offering plausible but different accounts of
visual word processing.
References
Anderson, J. R. (1978). Arguments concerning representations for mental
imagery. Psychological Review, 85, 249–277.
Bowers, J. S., & Turner, E. L. (2005). Masked priming is abstract in the
left and right visual fields. Brain and Language, 95, 414–422.
Bub, D., & Lewine, J. (1988). Different modes of word recognition in the
left and right visual fields. Brain and Language, 33, 161–188.
2
Another possibility for word length effects with familiar-format words
is less interesting, in that the main pattern of results could reflect statistical
pressures. Word processing is almost always more efficient following LH
presentations than following RH presentations, so statistical pressures
from the ceiling/floor could be responsible for attenuating word length
effects (that would otherwise be strong without the floor/ceiling) when
words are presented directly to the LH. Less such attenuation should
occur when words are presented directly to the RH.
C.J. Marsolek, R.G. Deason / Brain and Language 103 (2007) 304–307
Burgund, E. D., & Marsolek, C. J. (1997). Letter-case specific priming in
the right cerebral hemisphere with a form-specific perceptual identification task. Brain and Cognition, 35, 239–258.
Burgund, E. D., & Marsolek, C. J. (2000). Viewpoint-invariant and
viewpoint-dependent object recognition in dissociable neural subsystems. Psychonomic Bulletin & Review, 7, 480–489.
Burgund, E. D., Marsolek, C. J., & Luciana, M. (2003). Serotonin levels
influence patterns of repetition priming. Neuropsychology, 17, 161–170.
Christman, S. (1989). Perceptual characteristics in visual laterality
research. Brain and Cognition, 11, 238–257.
Chun, M. M., & Jiang, Y. (1998). Contextual cueing: implicit learning and
memory of visual context guides spatial attention. Cognitive Psychology, 36, 28–71.
Chun, M. M., & Jiang, Y. (1999). Top-down attentional guidance based on
implicit learning of visual covariation. Psychological Science, 10, 360–365.
Deason, R. G., & Marsolek, C. J. (2005). A critical boundary to the lefthemisphere advantage in visual word processing. Brain and Language,
92, 251–261.
Ellis, A. W. (2004). Length, formats, neighbours, hemispheres, and the
processing of words presented laterally or at fixation. Brain and
Language, 88, 355–366.
Ellis, A. W., Young, A. W., & Anderson, C. (1988). Modes of visual word
recognition in the left and right cerebral hemispheres. Brain and
Language, 35, 254–273.
Farah, M. J. (1991). Patterns of co-occurrence among the associative
agnosias: implications for visual object representation. Cognitive
Neuropsychology, 8, 1–19.
Jiang, Y., Song, J-H, & Rigas, A. (2005). High-capacity spatial contextual
memory. Psychonomic Bulletin & Review, 12, 524–529.
Koutstaal, W., Wagner, A. D., Rotte, M., Maril, A., Buckner, R. L., &
Schacter, D. L. (2001). Perceptual specificity in visual object priming:
functional magnetic resonance imaging evidence for a laterality
difference in fusiform cortex. Neuropsychologia, 39, 184–199.
Lavidor, M. (2002). An examination of the lateralized abstractive/form specific
model using MiXeD-CaSe primes. Brain and Cognition, 48, 413–417.
Lavidor, M., & Ellis, A. W. (2001). Mixed case effect in lateralised word
recognition. Brain and Cognition, 46, 192–195.
Lavidor, M., Ellis, A. W., & Pansky, A. (2002). Case alternation and
length effects in lateralized word recognition: studies of English and
Hebrew. Brain and Cognition, 50, 257–271.
Lavidor, M., Ellis, A. W., Shillcock, R., & Bland, T. (2001). Evaluating a
split processing model of visual word recognition: effect of word
length. Cognitive Brain Research, 12, 265–272.
307
Marsolek, C. J. (1999). Dissociable neural subsystems underlie abstract
and specific object recognition. Psychological Science, 10,
111–118.
Marsolek, C. J. (2003). What is priming and why? In J. S. Bowers & C. J.
Marsolek (Eds.), Rethinking implicit memory (pp. 41–64). Oxford:
Oxford University Press.
Marsolek, C. J. (2004). Abstractionist versus exemplar-based theories of
visual word priming: a subsystems resolution. The Quarterly Journal of
Experimental Psychology, 57, 1233–1259.
Marsolek, C. J., & Burgund, E. D. (1997). Computational analyses and
hemispheric asymmetries in visual-form recognition. In S. Christman
(Ed.), Cerebral asymmetries in sensory and perceptual processing
(pp. 125–158). Amsterdam: Elsevier.
Marsolek, C. J., & Burgund, E. D. (2003). Visual recognition and priming
of incomplete objects: the influence of stimulus and task demands. In J.
S. Bowers & C. J. Marsolek (Eds.), Rethinking implicit memory
(pp. 139–156). Oxford: Oxford University Press.
Marsolek, C. J., & Hudson, T. E. (1999). Task and stimulus demands
influence letter-case specific priming in the right cerebral hemisphere.
Laterality, 4, 127–147.
Marsolek, C. J., Kosslyn, S. M., & Squire, L. R. (1992). Form-specific
visual priming in the right cerebral hemisphere. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18,
492–508.
Marsolek, C. J., Nicholas, C. D., & Andresen, D. R. (2002). Interhemispheric communication of abstract and specific visual-form information. Neuropsychologia, 40, 1983–1999.
Marsolek, C. J., Schacter, D. L., & Nicholas, C. D. (1996). Form-specific
visual priming for new associations in the right cerebral hemisphere.
Memory & Cognition, 24, 539–556.
Marsolek, C. J., Squire, L. R., Kosslyn, S. M., & Lulenski, M. E. (1994).
Form-specific explicit and implicit memory in the right cerebral
hemisphere. Neuropsychology, 8, 588–597.
Price, C. J., & Devlin, J. T. (2003). The myth of the visual word form area.
Neuroimage, 19, 473–481.
Sergent, J., & Hellige, J. B. (1986). Role of input factors in visual-field
asymmetries. Brain and Cognition, 5, 174–199.
Vaidya, C. J., Gabrieli, J. D. E., Verfaellie, M., Fleischman, D., & Askari,
N. (1998). Font-specific priming following global amnesia and
occipital lobe damage. Neuropsychology, 12, 183–192.
Young, A. W., & Ellis, A. W. (1985). Different methods of lexical access
for words presented in the left and the right visual fields. Brian and
Language, 24, 364–371.