Vision Research 116 (2015) 210–218

Contents lists available at ScienceDirect

Vision Research
journal homepage: www.elsevier.com/locate/visres

Recent developments in a computational theory of visual attention (TVA)

Claus Bundesen ⇑, Signe Vangkilde, Anders Petersen
University of Copenhagen, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: This article reviews the foundations of the theory of visual attention (TVA) and describes recent develop-
Received 24 July 2014 ments in the theory. TVA is based on the principle of biased competition: All possible visual categoriza-
Received in revised form 10 November 2014 tions ascribing features to objects compete (race) to become encoded into visual short-term memory
Available online 29 November 2014
before it is filled up. Each of the possible categorizations is supported by sensory evidence, but the com-
petition is biased by multiplication with attentional weights (high weights on important objects) and per-
Keywords: ceptual biases (toward use of important categories). The way sensory evidence and attentional biases
Visual attention
interact is specified in the rate and weight equations of TVA, so TVA represents a mathematical formal-
TVA
ization of the biased competition principle. In addition to describing TVA as a psychological theory, we
present the neural interpretation of TVA, NTVA.
Ó 2014 Elsevier Ltd. All rights reserved.

0. Introduction 1. A computational theory of visual attention (TVA)

In this article, the foundations of the theory of visual attention, 1.1. Basic assumptions
TVA (Bundesen, 1990), are reviewed and recent developments in
the theory are described. TVA is based on the principle of biased In TVA, both visual recognition and selection of objects in the
competition (Desimone & Duncan, 1995): All possible visual cate- visual field consist in making visual categorizations of the form
gorizations ascribing features to objects compete (race) to become ‘‘object x has feature i’’ or, equivalently, ‘‘object x belongs to cate-
encoded into visual short-term memory before it is filled up. Each gory i.’’ A categorization is made (i.e., selected) when it is encoded
of the possible categorizations is supported by sensory evidence, into visual short-term memory (VSTM). If and when the categori-
but the competition is biased by multiplication with attentional zation is made, object x is said to be selected and also to be recog-
weights (high weights on important objects) and perceptual biases nized as a member of category i. Thus, selection and recognition
(toward use of important categories). The way sensory evidence are viewed as two aspects of the same process.
and attentional biases interact is specified in the rate and weight If and when a visual categorization of an object completes pro-
equations of TVA, so TVA represents a mathematical formalization cessing, the categorization enters VSTM if memory space for the
of the biased competition principle. categorization is available in VSTM. The storage capacity of VSTM
Bundesen (1990) showed how TVA can account for many psy- (parameter K) is normally assumed to be about four independent
chological findings (reaction times and error rates) in the field of objects (see, e.g., Luck & Vogel, 1997; Shibuya & Bundesen,
visual attention. More recently, Duncan et al. (1999) showed how 1988), however, K seems to vary not only between individuals
TVA provides a method to quantify individual variation in atten- but also on a trial-by-trial basis (see Dyrholm et al., 2011). Clearing
tional abilities, which has been used to study many normal and VSTM opens a race among all objects in the visual field to become
clinical populations. Later TVA was extended by Bundesen, encoded into VSTM. An object becomes encoded in VSTM if, and
Habekost, and Kyllingsbæk (2005, 2011) and Bundesen and only if, some categorization of the object becomes encoded in
Habekost (2008) into a neural theory of visual attention (NTVA), VSTM. Thus, each object x may be represented in the encoding race
which accounts for many effects of attention observed in firing by all possible categorizations of the object.
rates of single cells in the visual system of primates.

1.1.1. Rate equation

⇑ Corresponding author at: Department of Psychology, University of Copenhagen, By the rate equation of TVA, the rate, m(x, i), at which a particular
Øster Farimagsgade 2A, DK-1353 Copenhagen S, Denmark. visual categorization, ‘‘x belongs to i,’’ is encoded into VSTM is
E-mail address: bundesen@psy.ku.dk (C. Bundesen). given by a product of three terms:

http://dx.doi.org/10.1016/j.visres.2014.11.005
0042-6989/Ó 2014 Elsevier Ltd. All rights reserved.
 C. Bundesen et al. / Vision Research 116 (2015) 210–218 211

wx weight equations. In this sense, the theory yields a truly computa-

v ðx; iÞ ¼ gðx; iÞbi P : ð1Þ
z2S wz tional and mechanistic account of selective attention in vision.
The first term, g(x, i), is the strength of the sensory evidence that x
1.3. Applications to basic studies of human performance
belongs to category i. The second term, bi, is a perceptual decision
bias associated with category i (0 6 bi 6 1). The third term is the rel-
TVA has been applied to a broad range of findings on single-
ative attentional weight of object x—that is, the weight of object x,
stimulus recognition and selection from multielement displays
wx, divided by the sum of weights across all objects in the visual
both in normal subjects and in patient populations (see
field, S.
Bundesen & Habekost, 2008, 2014, for overviews). In addition,
the scope of the theory has been extended to other cognitive
1.1.2. Weight equation
domains such as memory and executive control.
The attentional weights in the rate equation of TVA are assumed
to be derived from pertinence values. Every visual category j is sup-
1.3.1. Single-stimulus recognition
posed to have a pertinence, pj, which is a nonnegative real number,
TVA provides a mathematical derivation of the biased-choice
reflecting the current importance of attending to objects that
model of Luce (1963) for single-stimulus recognition (see
belong to category j. By the traditional weight equation of TVA,
Bundesen, 1990, 1993). This model has been successful in explain-
the attentional weight of an object x in the visual field is
ing many experimental findings on effects of visual discriminabil-
X
wx ¼ gðx; jÞpj ; ð2Þ ity and bias (see, e.g., Townsend & Ashby, 1982; Townsend &
j2R Landon, 1982).
TVA also provides predictions of the time needed to recognize a
where R is the set of all visual categories, g(x, j) is the strength of the singly-presented simple stimulus. The v value on the left-hand side
sensory evidence that object x belongs to category j, and pj is the of Eq. (1) is a hazard rate. Thus, at any time t, the instantaneous
pertinence of category j. By Eq. (2), the attentional weight of an value of v(x, i) is the conditional probability density that the cate-
object is a weighted sum of pertinence values. The pertinence of a gorization that object x belongs to category i is encoded into VSTM
given category enters the sum with a weight equal to the strength at time t, given that the categorization has not been encoded before
of the sensory evidence that the object belongs to the category. time t, and given that storage space for the categorization is avail-
Based on partial-report experiments with color singletons, able in VSTM. This assumption implies that, if v(x, i) = 0 until time
Nordfang, Dyrholm, and Bundesen (2013) recently proposed a gen- t0, and m(x, i) is kept constant for times t > t0, the probability p(t) of
eralization of the weight equation such that making the categorization ‘‘x belongs to i’’ at or before time t is
X 0 exponentially distributed:
wx ¼ jx gðx; jÞpj ; ð2 Þ
j2R pðtÞ ¼ 1  exp½v ðx; iÞðt  t0 Þ:: ð3Þ
where jx is the strength of the local feature contrast of object x. Thus, When incorrect categorizations due to guessing or perceptual con-
Nordfang, Dyrholm, and Bundesen (2013) provided evidence that fusions among stimuli can be neglected, p(t) equals the probability
the attentional weight of an object is a product of both bottom-up of recognizing x as a member of i from a postmasked presentation
(feature contrast) and top-down (feature relevance) components. with an exposure duration of t, given that t0 equals the longest inef-
Note that, for displays in which the local feature contrast jx is the fective exposure duration.
same for all objects x, the change from Eq. (2) to Eq. (20 ) is immate- Recently, Dyrholm et al. (2011) provided evidence of a small
rial because the ratios between the attentional weights of different trial-by-trial variation in t0, which could be approximated by a nor-
objects remain the same. mal distribution. By this approximation, the probability of making
a particular categorization at or before time t is no longer exponen-
1.2. Mechanisms of selection tially distributed but follows an ex-Gaussian distribution (i.e., a
convolution of a normal and an exponential distribution; Luce,
TVA assumes two mechanisms of selection: a mechanism for 1986).
selection of objects and a mechanism for selection of categories. Eq. (3) has yielded close and useful fits to many empirical data,
Selection of objects is controlled by pertinence values and atten- including data from patient studies based on TVA (see Habekost &
tional weights. If selection of objects with a particular feature is Starrfelt, 2009; for a review). The equation presupposes that per-
desired, the pertinence of the feature in question should be high. ceptual confusions among stimuli can be neglected, but
The weight equation implies that when a feature has a high perti- Kyllingsbæk, Markussen, and Bundesen (2012) recently showed
nence, objects with the given feature get high attentional weights. how the time course of visual recognition of mutually confusable
By the rate equation, therefore, objects with the given feature are stimuli can be analyzed by use of TVA (also see Logan, 1996). Kyl-
likely to win the processing race and be encoded into VSTM. lingsbæk et al. proposed and tested a Poisson counter model of
Selection of categories is controlled by decision bias parame- visual recognition of briefly presented, mutually confusable single
ters. Whereas pertinence values determine which objects are stimuli in pure accuracy tasks (i.e., unspeeded tasks). The model
selected (filtering), decision biases determine how the objects are implies that during stimulus analysis, tentative categorizations
categorized (pigeonholing). If particular types of categorizations that stimulus x belongs to category i are made at a constant Poisson
are important, the bias values of the corresponding categories rate, v(x, i). The analysis is continued until the stimulus disappears,
should be high so that the desired types of categorizations are and the overt response is based on the categorization made the
likely to be made (cf. the rate equation). greatest number of times. This model provided a close fit to indi-
Coupled to a front-end system that supplies appropriate mea- vidual data on identification of digits and an apparently perfect
sures (g values) of the strength of the sensory evidence concerning fit to data on identification of Landolt rings.
the features of objects in its visual field, and given adequate set-
tings of pertinence (p) and bias (b) parameters, the specified sys- 1.3.2. Temporal attention in single-stimulus recognition
tem accomplishes both filtering and pigeonholing by an encoding Vangkilde, Coull, and Bundesen (2012) and Vangkilde, Petersen,
race between visual categorizations whose rate parameters are and Bundesen (2013) extended TVA to temporal attention (also see
determined by the simple algebraic operations of the rate and Matthias et al., 2009, 2010). Temporal expectations guide our
212 C. Bundesen et al. / Vision Research 116 (2015) 210–218

attention in time (see, e.g., Nobre, 2001, 2010; Nobre & Rohenkohl, outlined above was extended to multi-element displays.
2014). Beneficial effects of valid temporal expectations on motoric Vangkilde, Coull, and Bundesen (2012) and Vangkilde, Petersen,
responses have been demonstrated repeatedly over the past cen- and Bundesen (2013) explained the effect in terms of TVA by
tury (Woodrow, 1914; for reviews, see Los, 2010; Niemi & assuming that temporal expectations affect perception by chang-
Näätänen, 1981). Vangkilde, Coull, and Bundesen (2012) explored ing perceptual biases (values of b parameters). Specifically, a strong
effects of temporal expectation on visual processing speed in a expectation that a stimulus letter will appear at the next moment
cued single-stimulus recognition paradigm with briefly presented, should yield an increase in the b values of letter types, which
postmasked stimuli. Temporal expectations were manipulated by should speed the recognition of the stimulus letter if it appears
different hazard rate functions for the cue-stimulus foreperiod. when it is highly expected.
For example, in one experiment, the length of the foreperiod from
the cue to the stimulus was distributed exponentially. For each 1.3.3. Components of perceptual bias
block of trials, the participants knew from which of two exponen- Bundesen, Vangkilde, and Habekost (in press) proposed the
tial distributions with different hazard rates the foreperiods would general multiplicative hypothesis that the perceptual bias associ-
be drawn. The hazard rate could be either high (1.33 s1) or low ated with feature i (bi) in itself is a product of three terms:
(0.22 s1) corresponding to mean foreperiods of 750 ms and
bi ¼ Api ui ; ð4Þ
4500 ms, respectively. In either condition, the probability p(t) of
correct report as a function of the stimulus duration (t) was well where A is the level of alertness, pi is the subjective prior probability
described by Eq. (3), t0 being the threshold of conscious perception of being presented with feature i, and ui is the subjective impor-
(the longest ineffective exposure duration), and v(x, i) being the tance (‘‘utility’’) of identifying feature i.
speed of encoding into VSTM at times t > t0. As manipulated by By the rate equation of TVA, the rate of processing of a visual
the hazard rate, temporal expectation had no effect on the thresh- categorization that a given object has feature i is directly propor-
old of conscious perception but strong effect on the speed of sub- tional to bi. The multiplicative structure of Eq. (4) implies that
sequent encoding into VSTM. Averaged across participants, the bi = 0 if, and only if, either A = 0, pi = 0, or ui = 0. That bi = 0 if A = 0
speed of encoding was lowered by 30% in the low hazard rate con- entails the biologically very plausible implication that, if the level
dition. This effect was found even though no general decrease in of alertness is zero, no visual categorizations are made. The impli-
processing speed with time-on-task occurred. Thus, the effect cation that bi = 0 if the subjective prior probability of being pre-
was independent of the actual duration of the foreperiod on a given sented with feature i, pi, is zero, means that no objects are seen
trial, but depended entirely on expectation. These findings were as members of category i if the subject is absolutely certain that
extended in a parametric exploration of the effect of different lev- the category is empty. This would seem to characterize an ideal
els of temporal expectation (Vangkilde, Petersen, & Bundesen, observer. The implication that bi = 0 if the subjective importance
2013). In this study, the length of the cue-stimulus foreperiod of identifying feature i, ui, is zero, also would seem to be a charac-
was exponentially distributed with one of six hazard rates presum- teristic of an ideal observer. Humans may not be ideal observers,
ably giving rise to one of six levels of temporal expectation within a but the hypothetical description of performance given by Eq. (4)
particular block. Again, we found no effect of expectation on the at least provides a benchmark for the analysis of how alertness,
threshold of conscious perception but a strong and remarkably sys- prior knowledge and importance may directly affect human
tematic increase in perceptual processing speed with increasing performance.
expectation. Specifically, to a very good approximation processing
speed could be described by a linearly increasing function of the 1.3.4. Selection from multielement displays
logarithm of the level of expectancy given by the six different haz- TVA provides a mathematical derivation of the fixed-capacity
ard rates (see Fig. 1). Similar effects were recently found by independent race model (FIRM; Shibuya & Bundesen, 1988) for
Sørensen, Vangkilde, and Bundesen (2014) when the design selection from multielement displays consisting of homogeneous
elements (in the sense of Bundesen, 1990, p. 524). According to
FIRM, the processing of a stimulus display occurs in two stages.
At the first stage, an attentional weight is computed for each ele-
Ó Royal Society 2012

ment in the display. The weight of an element is a measure of

the strength of the sensory evidence that the element is a target.
At the second stage, the race between the elements takes place.
At this stage of the system, the total processing capacity is
assumed to be a constant, C elements per second. In the derivation
of FIRM from TVA, C is defined as the sum of all v values across all
perceptual categorizations of all elements in the visual field (cf.
Townsend & Ashby, 1983):
X XX
C¼ v ðxÞ ¼ v ðx; iÞ: ð5Þ
x2S x2S i2R

The processing capacity C is distributed across the display ele-

ments in proportion to their weights. Thus, every element in the
display is allocated a certain fraction of the total processing capac-
ity, and that fraction equals the attentional weight of the element
Fig. 1. Best linear fit to the mean estimated processing speed v (letters/s) for each divided by the sum of the attentional weights across all of the ele-
hazard rate condition plotted on a logarithmic scale. Conditions are: k = 0.1 s1 ments in the display. In many visual search tasks, the stimuli fall
(violet), k = 0.2 s1 (blue), k = 0.4 s1 (green), k = 0.8 s1 (yellow), k = 1.6 s1 into two categories: targets (i.e., items to be reported or otherwise
(orange), and k = 3.2 s1 (red). Error bars show standard errors of the means. The responded to) and distractors (items to be ignored), such that every
linear fit accounted for 99.4% of the variance of the estimated v values. From
‘‘Temporal Expectancy in the Context of a Theory of Visual Attention’’ by S.
target has approximately the same attentional weight as any other
Vangkilde, A. Petersen, and C. Bundesen, 2013, Philosophical Transactions of the Royal target, and every distractor has approximately the same atten-
Society, 368, 20130054, p. 8. tional weight as any other distractor. In such tasks, the efficiency
 C. Bundesen et al. / Vision Research 116 (2015) 210–218 213

(or inefficiency) of top-down selection can be measured by the the processing rate of a target and a distractor in each of the dis-
ratio, a, between the attentional weights of a distractor and a play types is given by C/(T + aD) and aC/(T + aD), respectively.
target: The close fit shown by the smooth curves was obtained with VSTM
capacity K at a value of 3.7 elements, total processing capacity C at
a ¼ wdistractor =wtarget : ð6Þ
49 elements/s, distractibility parameter a (the weight ratio of a dis-
The time it takes to encode an element into VSTM depends on the tractor to a target) at 0.40, and longest ineffective exposure dura-
amount of processing capacity allocated to the element. Specifically, tion t0 at 19 ms. Noninteger values of VSTM capacity K were
the encoding time is assumed to be exponentially distributed with a treated as probability mixtures such that, for example, a value of
rate parameter equal to the amount of processing capacity allocated 3.7 elements for K represented a mixture of values of 3 and 4 with
to the element. Encoding times for different elements are stochasti- a probability of .7 that K = 4 on a given trial.
cally independent, and the selected elements are those elements Many other well-established results have been closely fit by
whose encoding processes complete before (a) the stimulus presen- TVA. The results include findings of stochastic independence
tation terminates and (b) VSTM has been filled up. (Bundesen, Kyllingsbæk, & Larsen, 2003; Kyllingsbæk &
FIRM makes precise predictions of effects of variations in the Bundesen, 2007) in recognition of multiple features, effects of
exposure duration of the stimuli. Shibuya and Bundesen (1988) object integrality (Duncan, 1984), effects of number and spatial
tested such predictions in a comprehensive study of partial report position of targets in studies of divided attention (e.g., Posner,
of digits from mixtures of letters and digits with exposure dura- Nissen, & Ogden, 1978; Sperling, 1960, 1967), effects of selection
tions ranging from 10 ms up to 200 ms. The participants were criterion and the number of distractors in studies of focused atten-
instructed to report as many digits (targets) as possible from the tion (e.g., Bundesen & Pedersen, 1983; Treisman & Gelade, 1980;
stimulus display while ignoring the letters (distractors). Each dis- Treisman & Gormican, 1988), and effects of consistent practice in
play was terminated by a pattern mask. The results from one rep- search (Schneider & Fisk, 1982; also see Kyllingsbæk, Schneider,
resentative participant are illustrated in Fig. 2. For each & Bundesen, 2001).
combination of exposure duration, number of targets, and number
of distractors, the figure shows the probability distribution of the 1.3.5. Attentional dwell time
number of correctly reported targets. Each panel shows the results The majority of TVA-based studies have investigated attention
for a particular combination of the numbers of targets (T) and dis- using brief (e.g., 20–200 ms) exposures of simultaneously pre-
tractors (D). The top curve in the panel shows the probability of sented stimuli. However, Petersen, Kyllingsbæk, and Bundesen
reporting at least 1 target correct as a function of exposure dura- (2012) recently elaborated TVA in the temporal domain by pre-
tion, the second curve from the top shows the probability of senting a computational model (a theory of temporal visual atten-
reporting at least 2 targets correct, and so on. According to FIRM, tion; TTVA) to account for classical findings of attentional

Ó American Psychological Association 1988

Fig. 2. Relative frequency of scores of j or more correctly reported targets as a function of exposure duration with j, number of targets T, and number of distractors D as
parameters in partial report experiment of Shibuya and Bundesen (1988). Data are shown for Subject MP. Parameter j varies within panels; j is 1 (open circles), 2 (open
squares), 3 (solid squares), 4 (solid circles), or 5 (triangle). T and D vary among panels. Smooth curves represent a theoretical fit to the data by the FIRM model. For clarity,
observed frequencies less than .02 were omitted from the figure. Adapted from ‘‘Visual Selection From Multielement Displays: Measuring and Modeling Effects of Exposure
Duration,’’ by H. Shibuya and C. Bundesen, 1988, Journal of Experimental Psychology: Human Perception and Performance, 14, p. 595.
214 C. Bundesen et al. / Vision Research 116 (2015) 210–218

dynamics when two stimuli are presented in close temporal prox- 1.3.6. Relation to other cognitive domains
imity. Specifically, TTVA accounts for the attentional dwell time Gordon Logan has extended TVA to other cognitive domains
phenomenon, where the report of a masked target (T2) is severely than visual attention. First, Logan (1996) proposed a combination
impaired when it is presented with a delay of less than 500 ms of TVA with the Contour Detector (CODE) theory of perceptual
after a spatially separate masked target (T1) (see Duncan, Ward, grouping (gestalt formation) based on nearness that was proposed
& Shapiro, 1994; Ward, Duncan, & Shapiro, 1996). TTVA assumes by van Oeffelen and Vos (1982, 1983). The resulting theory—the
that a stimulus encoded and retained in VSTM takes up visual pro- CODE Theory of Visual Attention (CTVA)—integrates space-based
cessing resources, which could otherwise have been employed for and object-based approaches to visual attention. Thus, the theory
encoding subsequent stimuli into VSTM. These resources are explains many effects of spatial distance in visual attention (cf.
locked until the stimulus in VSTM has been recoded to a nonvisual Logan, 1996; Logan & Bundesen, 1996). CTVA has been formalized
(e.g., auditory, motoric, or amodal) format, which gives rise to the by Bundesen (1998).
long dwell time. Fig. 3 shows the observed and predicted perfor- Second, Logan and Gordon (2001) extended CTVA to ECTVA: a
mance with T1 and T2 for 3 individual subjects tested by theory of Executive Control of TVA in dual-task situations. This the-
Petersen, Kyllingsbæk, and Bundesen (2012) and for a group of ory accounts for crosstalk, set-switching cost, and concurrence
subjects reported in Duncan, Ward, and Shapiro (1994). costs in dual-task settings. The theory is based on the assumption
Petersen, Kyllingsbæk, and Bundesen (2013) used TTVA to that superordinate, executive processes coordinate and control
explain the well-known finding that removal of the mask of T1 subordinate processes by manipulating their parameters. TVA is
results in a faster recovery of T2 performance (Moore et al., used to describe the subordinate processes, and a task set is
1996; Raymond, Shapiro, & Arnell, 1992). This effect has previously defined as a set of TVA parameters (primarily gs, bs, and ps) that
been linked to a shorter dwell-time on unmasked targets, but suffices to configure a subordinate process to perform a task. Set
Petersen, Kyllingsbæk, and Bundesen (2013) found evidence that switching costs are explained in terms of the time it takes to
both targets and masks lock attentional resources such that masks change parameters and the number of parameters that need to
effectively function as distractors, suggesting that the fast recovery be changed.
of T2 performance may simply be due to T2 not having to compete Finally, by combining ECTVA with the exemplar-based random
with other objects for processing resources after the presentation walk model of categorization proposed by Nosofsky and Palmeri
(and decay) of T1. For stimuli presented in synchrony, TTVA (1997), Logan (2002) created an Instance Theory of Attention and
reduces to TVA, so TTVA explains all behavioral findings previously Memory (ITAM). The model of Nosofsky and Palmeri (1997) is itself
explained by TVA. a combination of Nosofsky’s (1986) generalized context model of
categorization and Logan’s (1988) instance theory of automaticity.
In ITAM, as in Logan’s (1988) instance theory of automaticity, a cat-

Ó Psychonomic Society 2012

Subject 1 Subject 2
1 1

0.8 0.8
eye

0.6 0.6
p , p or p
T2

0.4 0.4
T1

0.2 0.2

0 0
0 200 400 600 800 0 200 400 600 800

Subject 3 Duncan et al. (1994)

1 1

0.8 0.8
eye
p , p or p

0.6 0.6
T2

0.4 0.4 T1 (model)

T1

T2 (model)
T1 (data)
0.2 0.2 T2 (data)
Eye movements
0 0
0 200 400 600 800 0 200 400 600 800
SOA (ms) SOA (ms)

Fig. 3. Observed T1 (pT1, squares) and T2 (pT2, circles) performances as functions of stimulus onset asynchrony (SOA) for the 3 individual subjects tested by Petersen,
Kyllingsbæk, and Bundesen (2012) and for the group average reported in Duncan, Ward, and Shapiro (1994). Only trials without eye movements were analyzed. Bars indicate
the observed proportion of trials in which eye movements were registered, peye. Error bars indicate the standard deviations of pT1 and pT2 assuming that the responses from
the 3 individual subjects were approximately binomially distributed. Dashed lines (predicted T1 performance) and solid lines (predicted T2 performance) show least squares
fits of TTVA to the four data sets. From ‘‘Measuring and Modeling Attentional Dwell Time’’ by A. Petersen, S. Kyllingsbæk, and C. Bundesen, 2012, Psychonomic Bulletin &
Review, 19, p. 1030.
 C. Bundesen et al. / Vision Research 116 (2015) 210–218 215

egory is represented as a set of instances (individual examples or The weight equation of TVA describes how attentional weights
members of the category). Thus, ITAM is an integrated theory of are computed. The computation must occur before processing
attention, memory, and categorization. resources (cells) can be distributed in accordance with the weights.
Therefore, in NTVA, a typical perceptual cycle consists of two
waves: a wave of unselective processing, in which attentional
weights are computed, followed by a wave of selective processing,
2. A neural interpretation of TVA (NTVA)
once processing resources have been allocated in accordance with
the weights. During the first wave, cortical processing resources
2.1. Filtering and pigeonholing at the single-cell level
may be distributed at random (unselectively) across the visual
field. At the end of the first wave, an attentional weight has been
The neural theory of visual attention (NTVA; Bundesen,
computed for each object in the visual field and the weight has
Habekost, & Kyllingsbæk, 2005) provides a neural interpretation
been stored in a priority map. The weights are used for reallocation
of the two central equations of TVA, the rate and weight equations.
of attention (visual processing capacity) by dynamic remapping of
The equations jointly describe two mechanisms of attentional
receptive fields of cortical neurons. The remapping of receptive
selection: one for selection of objects and one for selection of cat-
fields makes the number of neurons allocated to an object increase
egories or, equivalently, features. NTVA specifies how the mecha-
with the attentional weight of the object. Thus, during the second
nisms work at the single-cell level: Filtering (selection of objects)
wave, cortical processing is selective in the sense that the amount
works by changing the number of cortical neurons in which an
of processing resources allocated to an object (the number of neu-
object is represented, whereas pigeonholing (selection of features)
rons that represent the properties of the object) varies with the
works by changing the rate of firing in cortical neurons coding for
attentional weight of the object. Because more processing
particular features (see Fig. 4). As explained below, in NTVA, the
resources are devoted to behaviorally important objects than to
total neural activation representing a visual categorization of the
less important ones, the important objects are processed faster,
form ‘‘object x has feature i’’ is directly proportional to both (a)
and are therefore more likely to become encoded into VSTM. Judg-
the number of neurons representing the categorization, which is
ing from data by Chelazzi et al. (1998, 2001), the second wave of
controlled by filtering, and (b) the level of activation of the individ-
processing may begin about 200 ms after the presentation of the
ual neurons representing the categorization, which is controlled by
stimulus array.
pigeonholing, and the rate equation simply expresses these direct
NTVA assumes that a typical neuron in the visual system is spe-
proportionalities.
cialized to represent a single feature (the feature preferred by the
Filtering makes the number of cells in which an object is repre-
neuron). The feature for which the neuron is specialized can be a
sented increase with the behavioral importance of the object. Thus,
more or less simple physical feature or a microfeature in a distrib-
in NTVA, visual processing occurs in parallel, with differential allo-
uted representation (cf. Hinton, McClelland, & Rumelhart, 1986).
cation of resources so that important objects become represented
NTVA also assumes that a visual neuron responds to the properties
in more cells than do less important objects. Specifically, in NTVA,
of only one object at any given time. In the first wave of processing
the probability that a cortical neuron represents a particular object
the object is selected at random among the objects in the neuron’s
in its classical receptive field equals the attentional weight of the
classical receptive field, but in the second wave of processing, the
object divided by the sum of the attentional weights across all
probability that the neuron represents a particular object equals
objects in the neuron’s receptive field (see Bundesen, Habekost, &
the attentional weight of the object divided by the sum of the
Kyllingsbæk, 2005, for a hypothetical neural network that may
attentional weights across all objects in the receptive field.
be used for the selection of objects).
In NTVA, the activation of a neuron by the appearance of an
object in its receptive field is defined as the increase in firing rate
above a baseline rate representing the spontaneous (undriven)
Ó American Psychological Association 2005

activity of the neuron. If the baseline rate is zero, the activation

equals the firing rate. In the suggested interpretation, the rate
Pigeonholing (rate of firing)

equation of TVA:

wx
v ðx; iÞ ¼ gðx; iÞbi P ;
z2S wz

describes the combined effects of filtering and pigeonholing on the

total activation of the population of neurons representing the cate-
gorization ‘‘object x has feature i.’’ The v value on the left-hand side
of the rate equation is the total activation of the neurons that rep-
resent the categorization at the level of processing in the brain
where objects compete for entrance into VSTM. At this level of pro-
cessing, the classical receptive fields of neurons are assumed to be
so large that each one covers nearly the entire visual field. Because
Filtering (number of cells) bi and

Fig. 4. Attentional selection in NTVA: Combined effects of filtering (selection of

wx
P ;
objects) and pigeonholing (selection of features) on the set of cortical spike trains z2S wz
representing a particular visual categorization of the form ‘‘object x has feature i.’’
The 4 conditions (quadrants) correspond to the factorial combinations of 2 levels of may only take on values up to 1, the rate equation implies that the
filtering (weak vs. strong support to object x) by 2 levels of pigeonholing (weak vs.
strength of the sensory evidence that x has feature i, g(x, i), is the
strong support to feature i). Filtering changes the number of cortical neurons in
which an object is represented. Pigeonholing changes the rate of firing of cortical highest possible value of v(x, i). Specifically, g(x, i) equals the total
neurons coding for a particular feature. From ‘‘A Neural Theory of Visual Attention: activation of the set of all neurons coding feature i when all of these
Bridging Cognition and Neurophysiology,’’ by C. Bundesen, T. Habekost, and S. cells represent object x (say, when x is the only object in the visual
Kyllingsbæk, 2005, Psychological Review, 112, p. 292. field) and when the featural bias in favor of i is maximal (i.e., bi = 1).
216 C. Bundesen et al. / Vision Research 116 (2015) 210–218

When the proportion of feature-i coding neurons representing including the frontal eye fields, anterior cingulate cortex, the claus-
object x: trum, and many thalamic nuclei. As also assumed in NTVA, Bisley
wx and Goldberg assume that the priority (attentional weight) of a
P ; stimulus reflects, in part, an evaluation of the subjective impor-
z2S wz
tance of the stimulus. The priorities in the map are thought to be
is smaller than 1, then the total activation representing the catego- used by the oculomotor system to target saccades and by the visual
rization ‘‘object x has feature i’’ is scaled down by multiplication system to guide visual attention. Bisley and Goldberg (2010) seem
with this factor on the right-hand side of the equation. The total to assume that visual attention to objects is a serial process, in
activation representing the categorization also varies in direct pro- which objects are processed one by one, but in NTVA this is only
portion to the level of activation of the individual neurons repre- true in special cases (see Bundesen, 1990, pp. 536–537).
senting the categorization. The bias parameter bi is a scale factor Finally, TVA is closely related to the normalization models of
that multiplies activations of all feature-i coding neurons, so the Reynolds and Heeger (2009; see also Moran & Desimone, 1985;
total activation representing the categorization ‘‘object x has feature Reynolds, Chelazzi, & Desimone, 1999) and Lee and Maunsell
i’’ is also directly proportional to bi. (2009). The model by Reynolds and Heeger was originally sug-
In summary, the neural interpretation of TVA’s rate equation gested as a way of implementing biased competition, which was
essentially reduces to the statement that the total activation repre- inspired by analyses of responses of cortical visual neurons to
senting the categorization ‘‘object x has feature i’’ is directly pro- simultaneously presented pairs of stimuli. In the normalization
portional to both the number of neurons representing the model, the excitatory neural activation caused by the stimuli (the
categorization (which is controlled by filtering) and the level of stimulus drive) is represented in a map that shows the activation
activation of the individual neurons representing the categoriza- as a function of both the locations of the receptive-field centers
tion (which is controlled by pigeonholing). of neurons and their feature preferences. The map of the stimulus
drive first becomes multiplied point-by-point with an attention
2.2. Related models of attention field and then becomes normalized by being divided point-by-
point with the suppressive drive to yield the output neural firing
The rate equation of TVA summarizes the way the strength rates. The suppressive drive is computed by convolving the
g(x, i) of the sensory evidence supporting the categorization ‘‘object point-by-point product of the stimulus drive and the attention
x has feature i’’ interacts with two types of biases: the relative field with a Gaussian kernel. Thus, the division by the suppressive
attentional weight of object x, drive has the effect that the attention-weighted stimulus drive
wx from a preferred stimulus is normalized with respect to the activity
P ; in other neurons that respond to the surrounding spatial and fea-
z2S wz
tural context. This model may be able to explain the way in which
and the perceptual bias associated with feature i, bi, in determining the shape of contrast-response functions for attended and non-
the total competitive strength v(x, i) of the given categorization. As attended stimuli change depending on both the size of the stimu-
explained in the introduction, the rate equation makes TVA a theory lus and the size of the attentional field. The effect of the normali-
of biased competition in the general sense of Desimone and Duncan zation is akin to the effect of using relative (i.e., normalized)
(1995). Of course, TVA is special in assuming two different types of instead of absolute attentional weights in the rate equation of
biases (one associated with objects and one with features) and also TVA. On the other hand, for a pair of adequate stimuli both of
in being formalized. which are presented within the classical receptive field of a
In addition to being a theory of biased competition, NTVA is a recorded cell, the normalization model implies that the response
feature similarity gain model in the sense of Treue and Martinez- (firing rate) of the cell is a weighted average of the responses
Trujillo (1999) and Martinez-Trujillo and Treue (2004). A gain obtained when either stimulus is presented alone (see Reynolds,
model of attention is a model in which attention works by multi- Chelazzi, & Desimone, 1999), whereas NTVA implies that the
plicative scaling of neuronal responses by a certain gain factor response to the stimulus pair is a probability mixture of the
(see McAdams & Maunsell, 1999). Treue and Martinez-Trujillo pro- responses obtained when the stimuli are presented alone (see
posed a fundamental principle, the feature similarity gain princi- Bundesen, Habekost, & Kyllingsbæk, 2005): With probability p,
ple, which says that the gain factor of a neuron increases with the cell responds as if only Stimulus 1 had been presented, and
increasing similarity between the sensory selectivity (the stimulus with probability 1  p, the cell responds as if only Stimulus 2 had
preferences) of the neuron and the currently attended features. been presented.
Thus, assuming that a neuron encodes those features it prefers,
attention to feature i should increase the responses of neurons 2.3. Functional anatomy
encoding feature i or encoding features similar to feature i. In terms
of TVA, the gain factor in question is the multiplicative perceptual NTVA is a fairly abstract neurocomputational model. As empha-
bias bi applied to neurons that are specialized for signalling feature sized by Bundesen, Habekost, and Kyllingsbæk (2005), the model
i. does not depend in a critical way on particular anatomical localiza-
NTVA also implies the saliency or priority map hypothesis pro- tions of the proposed computations. On the other hand, the model
posed by Itti and Koch (2000), Koch and Ullman (1985), Bisley suggests some ways in which visual computations may be distrib-
and Goldberg (2010), and others. By this hypothesis, the visual sys- uted across the human brain. One plausible distribution is given by
tem contains one or more maps of attentional weights or atten- the so-called thalamic version of NTVA, which is illustrated in
tional weight components. Specifically, according to a recent Fig. 5. For further discussions of the anatomy of visual attention,
version of the hypothesis, the lateral intraparietal area (LIP) acts see Bundesen, Habekost, and Kyllingsbæk (2005, 2011), Habekost
as a priority map in which objects are represented by activity pro- and Starrfelt (2009), and Gillebert et al. (2012).
portional to their behavioral priority. Bisley and Goldberg (2010)
point at evidence that the priority map combines bottom-up inputs 2.4. Applications to single-cell studies
from both the traditional dorsal and ventral streams of visual pro-
cessing, including areas V2, V3, V3a, MT, MST, V4, and IT, with top- Quantitative applications of NTVA to attentional effects
down inputs from a wide range of cortical and subcortical areas, observed in single-cell studies can be found in Bundesen,
 C. Bundesen et al. / Vision Research 116 (2015) 210–218 217

Ó American Psychological Association 2005

Bundesen, C. (1990). A theory of visual attention. Psychological Review,
97, 523–547.
Bundesen, C. (1993). The relationship between independent race models and Luce’s
choice axiom. Journal of Mathematical Psychology, 37, 446–471.
Bundesen, C. (1998). A computational theory of visual attention. Philosophical
Transactions of the Royal Society of London, Series B, 353, 1271–1281.
Bundesen, C., Vangkilde, S., & Habekost, T. (in press). Components of visual bias: A
multiplicative hypothesis. Annals of the New York Academy of Sciences.
Bundesen, C., & Habekost, T. (2008). Principles of visual attention: Linking mind and
brain. Oxford University Press.
Bundesen, C., Habekost, T., & Kyllingsbæk, S. (2005). A neural theory of visual
attention: Bridging cognition and neurophysiology. Psychological Review, 112,
291–328.
Bundesen, C., Habekost, T., & Kyllingsbæk, S. (2011). A neural theory of visual
attention and short-term memory. Neuropsychologia, 49, 1446–1457.
Bundesen, C., & Habekost, T. (2014). Theory of visual attention (TVA). In A. C. Nobre
& S. Kastner (Eds.), The Oxford handbook of attention (pp. 1095–1121). Oxford
University Press.
Bundesen, C., Kyllingsbæk, S., & Larsen, A. (2003). Independent encoding of colors
and shapes from two stimuli. Psychonomic Bulletin & Review, 10, 474–479.
Bundesen, C., & Pedersen, L. F. (1983). Color segregation and visual search. Attention,
Perception, & Psychophysics, 33, 487–493.
Fig. 5. Possible distribution of visual processing across the human brain. Visual Chelazzi, L., Duncan, J., Miller, E. K., & Desimone, R. (1998). Responses of neurons in
inferior temporal cortex during memory-guided visual search. Journal of
information from the eye enters the lateral geniculate nucleus (LGN) of the
Neurophysiology, 80, 2918–2940.
thalamus (1) and is transmitted to striate and extrastriate cortical areas where g
Chelazzi, L., Miller, E. K., Duncan, J., & Desimone, R. (2001). Responses of neurons in
values (strengths of evidence that objects at particular scales and positions have
macaque area V4 during memory-guided visual search. Cerebral Cortex, 11,
particular features) are computed (2). The g values are multiplied by p (pertinence) 761–772.
values, and the products are transmitted from the cortex to a priority map in the Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective visual attention.
pulvinar (Pul) nucleus of the thalamus, where the products are summed up as Annual Review of Neuroscience, 18, 193–222.
attentional weights of the stimulus objects (3). After the first (unselective) wave of Duncan, J. (1984). Selective attention and the organization of visual information.
processing, cortical processing capacity is redistributed by means of attentional Journal of Experimental Psychology: General, 113, 501–517.
weight signals (w) from the pulvinar to the cortex, so that during the second Duncan, J., Bundesen, C., Olson, A., Humphreys, G., Chavda, S., & Shibuya, H. (1999).
(selective) wave of processing, objects with high attentional weights are processed Systematic analysis of deficits in visual attention. Journal of Experimental
by many neurons (4). The resulting g values are multiplied by b (bias) values, and Psychology: General, 128, 450–478.
the products are transmitted from the cortex to a multiscale VSTM map of locations, Duncan, J., Ward, R., & Shapiro, K. (1994). Direct measurement of attentional dwell
which is tentatively localized in the thalamic reticular nucleus (TRN) (5). When the time in human vision. Nature, 369, 313–315.
VSTM map is initialized, objects in the visual field effectively start a race to become Dyrholm, M., Kyllingsbæk, S., Espeseth, T., & Bundesen, C. (2011). Generalizing
parametric models by introducing trial-by-trial parameter variability: The case
encoded into VSTM. In this race, each object is represented by all possible
of TVA. Journal of Mathematical Psychology, 55, 416–429.
categorizations of the object, and each possible categorization participates with an
Gillebert, C. R., Dyrholm, M., Vangkilde, S., Kyllingsbæk, S., Peeters, R., &
activation (v value) proportional to the corresponding g value multiplied by the Vandenberghe, R. (2012). Attentional priorities and access to short-term
corresponding b value. For the winners of the race, the TRN gates activation memory: Parietal interactions. Neuroimage, 62, 1551–1562.
representing a categorization back to some of those cells in LGN whose activation Habekost, T., & Starrfelt, R. (2009). Visual attention capacity: A review of TVA-based
supported the categorization (6). Thus activity in neurons representing winners of patient studies. Scandinavian Journal of Psychology, 50, 23–32.
the race is sustained by positive feedback. From ‘‘A Neural Theory of Visual Hinton, G. E., McClelland, J. L., & Rumelhart, D. E. (1986). Distributed
Attention: Bridging Cognition and Neurophysiology,’’ by C. Bundesen, T. Habekost, representations. In D. E. Rumelhart & J. L. McClelland (Eds.). Parallel
and S. Kyllingsbæk, 2005, Psychological Review, 112, p. 295. distributed processing (Vol. 1, pp. 77–109). Cambridge, MA: MIT Press.
Itti, L., & Koch, C. (2000). A saliency-based search mechanism for overt and covert
shifts of visual attention. Vision Research, 40, 1489–1506.
Habekost, and Kyllingsbæk (2005) and Bundesen and Habekost Koch, C., & Ullman, S. (1985). Shifts in selective visual attention: Towards the
(2008). The applications comprise attentional effects on the rate underlying neural circuitry. Human Neurobiology, 4, 219–227.
Kyllingsbæk, S., & Bundesen, C. (2007). Parallel processing in a multi-feature whole-
of firing with multiple stimuli in the receptive field of a neuron, report paradigm. Journal of Experimental Psychology: Human Perception and
effects on the rate of firing with a single stimulus in the receptive Performance, 33, 64–82.
field, and effects on baseline firing. Neural filtering is viewed as Kyllingsbæk, S., Markussen, B., & Bundesen, C. (2012). Testing a Poisson counter
model for visual identification of briefly presented, mutually confusable single
dynamic remapping of receptive fields, changing the number of
stimuli in pure accuracy tasks. Journal of Experimental Psychology: Human
neurons representing perceptual categorizations of a given type Perception and Performance, 38, 628–642.
of objects. Neural pigeonholing is viewed as a mechanism that Kyllingsbæk, S., Schneider, W. X., & Bundesen, C. (2001). Automatic attraction of
selects the set of neurons that prefer a given feature for a multipli- attention to former targets in visual displays of letters. Attention, Perception, &
Psychophysics, 63, 85–98.
cative increase or decrease in activation. Lee, J., & Maunsell, J. H. R. (2009). A normalization model of attentional modulation
of single unit responses. PLoS One, 4(2), e4651.
Logan, G. D. (1988). Toward an instance theory of automatization. Psychological
3. Final remark Review, 95, 492–527.
Logan, G. D. (1996). The CODE theory of visual attention: An integration of space-
In this article, we have reviewed the foundations of a formal based and object-based attention. Psychological Review, 103, 603–649.
Logan, G. D. (2002). An instance theory of attention and memory. Psychological
computational theory of visual attention, TVA (Bundesen, 1990), Review, 109, 376–400.
and presented some recent developments in the theory. We have Logan, G. D., & Bundesen, C. (1996). Spatial effects in the partial report paradigm: A
also summarized a neural interpretation of TVA, NTVA challenge for theories of visual spatial attention. In D. L. Medin (Ed.). The
psychology of learning and motivation (Vol. 35, pp. 243–282). San Diego, CA:
(Bundesen, Habekost, & Kyllingsbæk, 2005, 2011). Academic Press.
Logan, G. D., & Gordon, R. D. (2001). Executive control of visual attention in dual-
task situations. Psychological Review, 108, 393–434.
Acknowledgment
Los, S. A. (2010). Foreperiod and sequential effects: Theory and data. In A. C. Nobre &
J. T. Coull (Eds.), Attention and time (1st ed., pp. 289–302). Oxford University
This work was supported by a grant from the University of Press.
Copenhagen Programme of Excellence to Claus Bundesen. Luce, R. D. (1963). Detection and recognition. In R. D. Luce, R. R. Bush, & E. Galanter
(Eds.). Handbook of mathematical psychology (Vol. 1, pp. 103–189). New York:
Wiley.
References Luce, R. D. (1986). Response times: Their role in inferring elementary mental
organization. New York: Oxford University Press.
Luck, S. J., & Vogel, E. K. (1997). The capacity of visual working memory for features
Bisley, J. W., & Goldberg, M. E. (2010). Attention, intention, and priority in the
and conjunctions. Nature, 390, 279–281.
parietal lobe. Annual Review of Neuroscience, 33, 1–21.
218 C. Bundesen et al. / Vision Research 116 (2015) 210–218

Martinez-Trujillo, J. C., & Treue, S. (2004). Feature-based attention increases the Reynolds, J. H., Chelazzi, L., & Desimone, R. (1999). Competitive mechanisms
selectivity of population responses in primate visual cortex. Current Biology, 14, subserve attention in macaque areas V2 and V4. Journal of Neuroscience, 19,
744–751. 1736–1753.
Matthias, E., Bublak, P., Costa, A., Müller, H. J., Schneider, W. X., & Finke, K. (2009). Reynolds, J. H., & Heeger, D. J. (2009). The normalization model of attention. Neuron,
Attentional and sensory effects of lowered levels of intrinsic alertness. 61, 168–185.
Neuropsychologia, 47, 3255–3264. Schneider, W., & Fisk, A. D. (1982). Degree of consistent training: Improvements in
Matthias, E., Bublak, P., Müller, H. J., Schneider, W. X., Krummenacher, J., & Finke, K. search performance and automatic process development. Attention, Perception,
(2010). The influence of alertness on spatial and nonspatial components of & Psychophysics, 31, 160–168.
visual attention. Journal of Experimental Psychology: Human Perception and Shibuya, H., & Bundesen, C. (1988). Visual selection from multielement displays:
Performance, 36, 38–56. Measuring and modeling effects of exposure duration. Journal of Experimental
McAdams, C. J., & Maunsell, J. H. R. (1999). Effects of attention on orientation-tuning Psychology: Human Perception and Performance, 14, 591–600.
functions of single neurons in macaque cortical area V4. Journal of Neuroscience, Sørensen, T. A., Vangkilde, S., & Bundesen, C. (2014). Components of attention
19, 431–441. modulated by temporal expectation. Journal of Experimental Psychology:
Moore, C. M., Egeth, H., Berglan, L. R., & Luck, S. J. (1996). Are attentional dwell times Learning, Memory, and Cognition. http://dx.doi.org/10.1037/a0037268.
inconsistent with serial visual search? Psychonomic Bulletin & Review, 3, Sperling, G. (1960). The information available in brief visual presentations.
360–365. Psychological Monographs, 74(11, Whole No. 498).
Moran, J., & Desimone, R. (1985). Selective attention gates visual processing in the Sperling, G. (1967). Successive approximations to a model for short-term memory.
extrastriate cortex. Science, 229, 782–784. Acta Psychologica, 27, 285–292.
Niemi, P., & Näätänen, R. (1981). Foreperiod and simple reaction time. Psychological Townsend, J. T., & Ashby, F. G. (1982). Experimental test of contemporary
Bulletin, 89, 133–162. mathematical models of visual letter recognition. Journal of Experimental
Nobre, A. C. (2001). Orienting attention to instants in time. Neuropsychologia, 39, Psychology: Human Perception and Performance, 8, 834–864.
1317–1328. Townsend, J., & Ashby, F. G. (1983). The stochastic modeling of elementary
Nobre, A. C. (2010). How can temporal expectations bias perception and action? In psychological processes. Cambridge, UK: Cambridge University Press.
A. C. Nobre & J. T. Coull (Eds.), Attention and time (1st ed., pp. 371–392). Oxford Townsend, J. T., & Landon, D. E. (1982). An experimental and theoretical
University Press. investigation of the constant-ratio rule and other models of visual letter
Nobre, A. C., & Rohenkohl, G. (2014). Time for the fourth dimension in attention. In confusion. Journal of Mathematical Psychology, 25, 119–162.
A. C. Nobre & S. Kastner (Eds.), The Oxford handbook of attention (pp. 676–723). Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention.
Oxford University Press. Cognitive Psychology, 12, 97–136.
Nordfang, M., Dyrholm, M., & Bundesen, C. (2013). Identifying bottom-up and top- Treisman, A. M., & Gormican, S. (1988). Feature analysis in early vision: Evidence
down components of attentional weight by experimental analysis and from search asymmetries. Psychological Review, 95, 15–48.
computational modeling. Journal of Experimental Psychology: General, 142, Treue, S., & Martinez-Trujillo, J. C. M. (1999). Feature-based attention influences
510–535. motion processing gain in macaque visual cortex. Nature, 399, 575–579.
Nosofsky, R. M. (1986). Attention, similarity, and the identification– van Oeffelen, M. P., & Vos, P. G. (1982). Configurational effects on the enumeration
categorization relationship. Journal of Experimental Psychology: General, 115, of dots: Counting by groups. Memory & Cognition, 10, 396–404.
39–57. van Oeffelen, M. P., & Vos, P. G. (1983). An algorithm for pattern description on the
Nosofsky, R. M., & Palmeri, T. J. (1997). An exemplar-based random walk model of level of relative proximity. Pattern Recognition, 16, 341–348.
speeded classification. Psychological Review, 104, 266–300. Vangkilde, S., Coull, J. T., & Bundesen, C. (2012). Great expectations: Temporal
Petersen, A., Kyllingsbæk, S., & Bundesen, C. (2012). Measuring and expectation modulates perceptual processing speed. Journal of Experimental
modeling attentional dwell time. Psychonomic Bulletin & Review, 19, Psychology: Human Perception and Performance, 38, 1183–1191.
1029–1046. Vangkilde, S., Petersen, A., & Bundesen, C. (2013). Temporal expectancy in the
Petersen, A., Kyllingsbæk, S., & Bundesen, C. (2013). Attention dwells on both targets context of a theory of visual attention. Philosophical Transactions of the Royal
and masks. Journal of Vision, 13, 1–12. Society of London, Series B, 368, 20130054. http://dx.doi.org/10.1098/
Posner, M. I., Nissen, M. J., & Ogden, W. C. (1978). Attended and unattended rstb.2013.0054.
processing modes: The role of set for spatial location. In H. L. Pick & I. J. Ward, R., Duncan, J., & Shapiro, K. (1996). The slow time-course of visual attention.
Saltzman (Eds.), Modes of perceiving and processing information (pp. 137–157). Cognitive Psychology, 30, 79–109.
Hillsdale, NJ: Erlbaum. Woodrow, H. (1914). The measurement of attention. Psychological Monographs,
Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visual 17(5, Whole No. 76), 1–158.
processing in an RSVP task: An attentional blink? Journal of Experimental
Psychology: Human Perception and Performance, 18, 849–860.