Brain Structure and Function (2023) 228:1095–1105

https://doi.org/10.1007/s00429-022-02587-7

REVIEW

Do we understand the prefrontal cortex?

Richard E. Passingham1 · Hakwan Lau2

Received: 2 October 2022 / Accepted: 17 October 2022 / Published online: 8 November 2022
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2022

Abstract
Many suggestions have been made as to the functions of the prefrontal (PF) cortex. However, they involve labelling areas using
psychological terminology. But what we need to know is how the PF cortex operates. We argue that understanding this must start
with describing the flow of information. We illustrate this argument by considering three PF areas. Each has a unique pattern of
inputs and outputs, and we suggest that the implication is that each performs a unique transformation from the inputs to the outputs.
The caudal PF cortex transforms input that is maintained by attention or short-term memory into the target of the appropriate eye
movement. The mid-dorsal PF cortex transforms input concerning the order of objects or actions into the target of the appropriate
eye and hand movements, thus supporting sequences of action. The ventral PF cortex transforms input concerning an object or
sound into prospective activity that encodes the associated object or sound. However, it is important to appreciate that the mid-
dorsal and ventral PF cortex are specialized for encoding abstract transformations, irrespective of the specific actions or objects.
The advantage is that this enables generalization to novel problems that have the same underlying logic. We account for the dif-
ference between fast learning and slow learning in this way. The human brain has co-opted these mechanisms so as to support
intelligence. Non-verbal tests of IQ typically use sequences of letters, numbers or designs. These test the ability to understand the
abstract rules that apply. Here the activations lie in the mid-dorsal PF cortex. Verbal tests typically assess the ability to understand
semantic associations. These can be presented either in pictorial or verbal form. Here the activations lie in the ventral PF cortex.

Keywords Prefrontal cortex · Connectivity fingerprint · Functional anatomy

Abbreviations Introduction
FEF Frontal eye field
LIP Lateral intraparietal cortex There has been a prolonged debate about the function of the
PF Prefrontal prefrontal (PF) cortex.
PFG Intermedial area in inferior parietal cortex Sixty years ago, Teuber (1964) thought it to be a ‘riddle’.
PG Posterior part of the inferior parietal cortex Then Goldman-Rakic (1995), Goldman-Rakic 1996) pro-
PreSMA Presupplementary motor cortex posed that it provided a mechanism for working memory,
SII Secondary somatosensory area and there are those who still support this view (Constanti-
STS Superior temporal sulcus nidis and Qi 2018; Miller et al. 2018). Others have suggested
TE Anterior inferotemporal cortex a role in response selection (Passingham 1993), conditional
TEO Posterior inferotemporal cortex learning on the basis of single events (Passingham and Wise
VIP Ventral intraparietal cortex 2012) or planning (Shallice 1982). And more generally it is
V4 Fourth area in the hierarchy of the ventral often said that the PF cortex has executive functions and acts
visual system in cognitive control (Menon and D'Esposito 2022).
Duncan (2010) has taken a different tack, arguing that
any proposals are misguided if they were based on a narrow
* Hakwan Lau
hakwan.lau@riken.jp choice of behavioural tasks. He suggests that in fact the PF
cortex is engaged by (Fedorenko et al. 2013) and can learn
Richard E. Passingham
dick.passingham@psy.ox.ac.uk (Duncan 2001) a very wide variety of tasks. He therefore
suggests that it should be regarded as a ‘multiple demand
1
Department of Experimental Psychology, University system’.
of Oxford, Oxford, UK
2
RIKEN Center for Brain Science, Wako, Japan

13
Vol.:(0123456789)
1096 Brain Structure and Function (2023) 228:1095–1105

A recent book has reviewed the literature at length (Pass- the goal of the action by which we mean the target of
ingham 2021) with the bold title ‘Understanding the Prefron- either an eye movement or a hand movement (Passingham
tal Cortex’. The present paper outlines one of the themes of and Wise 2012). For example, when a macaque monkey
the book, suggesting general principles on how the problem moves a cursor on a visual maze, the cell activity in the
should be approached. This is based on considering the ana- mid-dorsal PF cortex relates to the goal, not to the direc-
tomical properties of the brain. The conclusions explain how tion in which the joystick has to be turned for the cursor
the different account of the functions of the PF cortex out- to reach that goal (Mushiake et al., 2006).
lined above can be accounted for if this anatomical approach The final implication is that the fact that the PF cortex
is adopted. lies at the top of the sensory and action hierarchy means
that it can both select the sensory cues that are relevant for
the current task as well as selecting the goal of the action
Transformation for the motor system. As Shipp (2005) pointed out the
cortical feedback pathways are similar to the output path-
In cognitive neuroscience, functions are often described in ways—and unlike the feedforward pathways—in that they
psychological terms, rather as in phrenology. There are two do not project to layer 4 (Shipp 2005). The selection occurs
limitations. The first is that these terms were introduced to via the mechanism of attentional enhancement. Thus, the
explain performance on laboratory tasks, and these have PF cortex influences and enhances the activity in task-rele-
often taken no account of the ecological conditions in which vant sensory association areas via feedback pathways (Gee
the primates evolved. The second is that the aim of neurosci- et al. 2008; Heinen et al. 2014; Siegel et al. 2015). And
ence is not to simply label the brain, but to understand how similarly, it influences and enhances activity in the premo-
it works. tor association areas including the presupplementary motor
This involves describing the transformation that is per- cortex (preSMA) and anterior cingulate cortex via output
formed from the inputs of an area to its outputs. Each area pathways (Rowe et al. 2002a, b; Westendorff et al. 2016).
has a unique overall pattern of inputs and outputs (Passing-
ham et al. 2002). However, the transformation depends not
only on the extrinsic connectivity of the area but also on its
intrinsic connectivity, that is the internal wiring (Passingham Three subdivisions of the PF cortex
et al. 2002). Rather than describing the flow of information
in terms of black box models (Broadbent 1958), the aim is Since each area of the brain has a different overall pattern
to produce an account that describes the anatomical inputs of inputs and outputs (Passingham et al. 2002), the trans-
and thus the information that an area receives, specifying formation will differ from one area to another within the
how that is converted into the outputs and thus the influence PF cortex. To illustrate the transformations, the discussion
that the area exerts within the system. In the conclusions we that follows considers three broad regions of the PF cortex,
point to the need to produce models of how this conversion the caudal, mid-dorsal, and ventral PF cortex. These were
occurs, given the internal wiring of the area. chosen because the debate as described above has been pri-
There is an important advantage in viewing the problem in marily about the function of these areas.
terms of the transformation from inputs to outputs or compu- Figure 1 shows these areas for the brain of a macaque
tation. This is that it avoids positing a homunculus that acts as monkey.
a controller. Baddeley’s working memory model has several In the macaque monkey and human brain these subdivi-
components, a short-term buffer and a central executive for sions include the following areas as defined by cytoarchitec-
example. But as he admitted he could be criticised for treat- ture (Petrides and Pandya 1999) in the macaque and human
ing the central executive as a homunculus (Baddeley 1996). brains:
Yet the PF cortex is no homunculus. It lies at the top of
the sensory hierarchy and at the top of the action hierarchy (1) The caudal PF cortex (navy blue) includes areas 8B,
(Passingham and Wise 2012). There are three implications. 8Ad, 8Av and 45B.
The first is that the PF cortex is in a position to transform (2) The mid-dorsal PF cortex (pink) includes areas 9/46
sensory inputs, whether current or memory, into action. For and 46.
example, the inferotemporal cortex sends connections to the (3) The ventral PF cortex (light blue) includes areas 45B
ventral PF cortex (Webster et al. 1994), not the premotor and 47/12.
cortex, and this means that it is the PF cortex that transforms
visual information about the context into action. These areas have also been delineated using connectiv-
The second implication is that the PF cortex does not ity in the macaque and human brains (Neubert et al. 2014;
actually specify the movement itself. Instead, it specifies Sallet et al. 2013).

13
Brain Structure and Function (2023) 228:1095–1105 1097

Fig. 1  The caudal, mid-dorsal and ventral PF cortex shown on the

brain of a macaque monkey. The figure also shows the orbital PF cor-
tex. Adapted from Price and Drevets (2010)
Fig. 2  Population vectors showing the gradual transformation during
a delay period from the input location to the target location at 90°.
Caudal PF cortex Adapted from Takeda and Funahashi (2004)

The caudal PF cortex transforms information in short-term

memory into the goal of a saccade (Passingham 2021). Area the ability to remember items. Instead, they cause errors
8Ad receives inputs concerning location from the lateral and in the accuracy of the eye movement (Mackey et al. 2016;
ventral intraparietal cortex (LIP, VIP) and posterior inferior Mackey and Curtis 2017) or the focus of attention (Keller
parietal cortex (PG) (Stanton et al. 1995). Area 8Av receives et al. 2008).
inputs concerning objects from area V4 and the anterior and Thus, the caudal PF cortex transforms the visual input
posterior inferotemporal cortex (TE, TEO) (Stanton et al. from posterior areas into the goal of the response. This trans-
1995). In turn area 8 can influence eye movements via out- formation was illustrated in a study by Takeda and Funa-
puts to the superior colliculus both directly (Kunzle et al. hashi (2004) in which they trained macaque monkeys on a
1976) and indirectly via the medial caudate and substantia task in which they had to remember a location but make a
nigra, pars reticulata (Hedreen and DeLong 1991). saccade to a location at 90°. Population vectors were cal-
There are cells that respond to the visual stimuli (V cells), culated for the cells in the caudal PF cortex, including the
cells that respond both to the visual stimulus and eye move- posterior part of area 9/46. As shown in Fig. 2 there was a
ment (VM cells) and cells that respond to the eye move- change in the direction of these vectors during the delay,
ment alone (M cells) (Schall 1991), though each of these indicating the gradual transformation of the memory of the
three types can be further subdivided into more specialized visual input into the goal of the saccade.
cells (Lowe and Schall 2018). It is the existence of cells
with these properties that explains how the area can trans-
form visual inputs concerning location into the goal of the Mid‑dorsal PF cortex
saccade.
There is activity in macaque monkeys (Blacker and The mid-dorsal PF cortex differs in that it is specialized for
Courtney 2016; Chafee and Goldman-Rakic 1998) and transforming information concerning the order of items into
humans (Rowe et al. 2000; Sala et al. 2003) in the caudal the goals of the action and generates sequences of goals
PF cortex during the delay period on a short-term memory (Passingham 2021). Area 9/46 has inputs from LIP and area
task. But it is persistent activity in posterior areas that is 8A, and outputs to the frontal eye field (FEF) and the supe-
critical for the maintenance of the items, whether in short- rior colliculus. These connections mean that it can transform
term memory (Mackey and Curtis 2017; Sreenivasan et al. a location into the goal of the next eye movement (Borra
2014) or as the focus for attention (Colby et al. 1996; Jerde et al. 2014; Borra et al. 2019; Gerbella et al. 2013).
et al. 2012). Lesions whether in macaque monkeys (Mackey By contrast, area 46 has inputs from the intermediate
and Curtis 2017; Tsujimoto and Postle 2012) or human sub- part of the inferior parietal cortex (PFG) and the secondary
jects (Ferreira et al. 1998; Mackey et al. 2016) fail to abolish somatic area (SII); and outputs to the premotor and preSMA.

13
1098 Brain Structure and Function (2023) 228:1095–1105

These connections mean that it can transform an action as between a cell coding for a subgoal and another coding for
performed or viewed into the goal of the next hand move- the goal during the transformation from one goal to the next
ment (Borra et al. 2019; Gerbella et al. 2013). one in time (Sakamoto et al. 2008).
The original evidence that order was critical came from
a study of macaque monkeys with lesions in the dorsal PF
cortex; they made errors on reporting the order of items with Ventral PF cortex
which they have been presented (Petrides 1991). More recent
neurophysiological and fMRI studies have provided clues The ventral PF cortex differs yet again in that it associates
as to the mechanism. There are cells that are active while one member of a pair with the other (Passingham 2021).
monkeys view a sequence (Rodriguez et al 2022), or com- Area 47/12 has visual and tactile inputs concerning objects
mit a sequence of items to short-term memory (Lundqvist (Petrides and Pandya 2002; Saleem et al. 2008) whereas
et al. 2018; Miller et al. 2018). There is also cell activity area 45A also has auditory ones from the superior temporal
when they report the order of items in a sequence (Warden cortex and Broca’s 45B (Petrides and Pandya, 2002). The
and Miller 2010), and while they plan a sequence of goals ventral PF cortex can influence choice via indirect projec-
(Averbeck et al. 2006; Mushiake et al. 2006). tions to the premotor cortex via the preSMA (Takahara et al.
Similarly, there is activation while human subjects com- 2012). These connections mean that the ventral PF cortex
mit a sequence of items to short-term memory (Amiez can transform information about objects or sounds into the
and Petrides 2007), maintain the order before repeating it choice of objects or sounds.
(Pochon et al. 2001), and generate a sequence of actions on To put it another way, it can learn to associate items. And
a planning task (Rowe et al. 2001; Spreng et al. 2010; Unter- unlike the dorsal PF cortex it does so irrespective of the
rainer et al. 2005) or generate a random sequence of goals by order in which they are presented. Even though a delay is
monitoring what they have just done (Phillips et al. 2018). sometimes presented between each member of the associated
Furthermore, lesions that include the mid-dorsal PF cor- pair, this is not necessary: they can be presented simultane-
tex severely impair the ability of patients to repeat the order ously (Bussey et al. 2001; Rushworth et al. 1997; Wang et al.
in which items were presented (Ferreira et al. 1998). This is 2000). In other words, the animal can choose the correct
not due to a failure to maintain the items in memory since item while the sample item is still present with no require-
the patients can recognize the order if testing on a matching ment for learning a sequence.
design (Ferreira et al. 1998; Pochon et al. 2001). In other Figure 3 illustrates the activity of cell populations while
words, the problem is one of transforming the order of the macaque monkeys associate two pictures (Rainer et al.
items into a sequence of goals. 1999). After the presentation of picture A, there is prospec-
Sequences require the transformation from one item or tive activity that encodes picture B, and this increases up to
goal to the next one in time. Tsujimoto et al. (2008) found the time that picture B is presented. In other words, there
pairs of cells in the mid-dorsal PF cortex of macaque mon- is a transformation between the input for picture A and the
keys in which one coded for the previous goal and the other output of specifying picture B as the goal.
for the current goal. At the time that the animal chose the Inactivation and lesion studies show that the ventral PF
current goal there was transient synchrony between the cortex is critical for learning associations. Area 45A is spe-
activity of the cells representing the previous and current cialized for the association between faces and vocalizations
goal. On a planning task there is also maximal synchrony (Hwang and Romanski 2015). The ventral PF cortex can

Fig. 3  The filled box shows the

time at which the input picture
was presented. The ascend-
ing line shows the prospective
population activity encoding
the target picture. Adapted from
Rainer et al. (1999)

13
Brain Structure and Function (2023) 228:1095–1105 1099

also associate items that are identical as in visual matching independently of the specific actions (Shima et al. 2007).
or are totally different, for example a visual cue with a spa- Figure 4 shows the population activity for cells firing in
tial target (Bussey et al. 2001; Wang et al. 2000). Freedman advance of sequences of four actions. The upper traces
et al. (2002) also found cells in the ventral PF cortex that show the activity for actions in which the animal had to
associated stimuli that were morphed so as to be similar but alternate what it did. The same cell populations did not
not identical, such as different cats or different dogs. These fire for sequences with a different structure, for example
cell populations act to classify objects (Cromer et al. 2010). repetition of actions.
However, the ventral PF contributes mainly to the learn- There are also cells in the ventral as well as the mid-
ing of new associate pairs because it has access to infor- dorsal cortex that code for abstract matching rules. These
mation about rewards via its connections with the orbital encoding whether the rule is matching or non-matching,
PF cortex (Petrides and Pandya 2002). There are cells that irrespective of the actual objects presented (Wallis et al.
encode these associations (Andreau and Funahashi 2011). 2001).
It appears to be less critical for the retrieval of overlearned The advantage of learning abstract rules is that it means
associates (Boettiger and D'Esposito 2005; Wang et al. that anthropoid primates can show transfer or generalize
2000). It is true that Bussey et al. (2001) reported an impair- from solving one problem to solving others if the underly-
ment in learning new visual-spatial associates, in spite of ing logic is the same (Harlow and Warren 1952) or related
training before surgery. One reason could be that the lesions (Warren 1974). In other words, they can develop a learn-
in this study included not only the ventral but also the orbital ing set. Rather than starting from scratch on each problem
PF cortex. (slow learning) they can show rapid improvement across
problems (fast learning).
The distinction between slow and fast learning was
Abstract rules and fast learning illustrated by a study in which macaque monkeys learned
visuo-spatial associations (Bussey et al. 2001). The ani-
Though the PF cortex transforms from sensory input to mals with ventral and orbital PF lesions could learn visuo-
action output, this cannot be the full story because rodents spatial associations over very many sessions, presumably
and other mammals can do this. Yet they lack a granular PF because there was still a visual input to the basal ganglia
cortex including the lateral PF areas described above (Preuss (Hoshi 2013). However, they showed no improvement
2007; Wise 2017). This raises the question as to what the at all on new visuo-spatial problems within a session.
granular PF cortex adds. Improvement within a session depends on learning the
The answer is that it supports the ability to learn abstract abstract rules ‘repeat-stay’ and ‘change-shift’, and Geno-
rules or transformations. It is in a position to do this because vesio and Wise (2008) have described subpopulations of
it lies at the top of the sensory and action hierarchies, and it cells in the PF that code for these abstract strategies.
is this hierarchical organization (Markov and Kennedy 2013) Fast learning involves the development of statistical
that allows the abstraction of specific features. For example, assumptions or priors (Botvinick et al. 2019; Costa et al.
the projections from the inferotemporal cortex to the perirhi- 2015). Wang et al. (2018) take the example of a two-bandit
nal cortex support view invariance of objects (Buckley and problem and suggest that, whereas the basal ganglia can
Gaffan 1998; Wang et al. 1998). Cadieu et al. (2007) pro- support slow learning via reinforcement, the dopamine
duced a model that was able to achieve invariance for posi- system also trains up the PF cortex so that it can operate
tion or shape by combining units in a hierarchical fashion. independently as an algorithm that supports fast learning.
There are cells in the mid-dorsal PF cortex that code They describe this as ‘meta-learning’.
for abstract sequences, that is the structure of sequences

Fig. 4  The green lines show

cell populations encoding the
abstract structure of sequences
in which the actions alternated.
The response to sequences with
a different abstract structure is
labelled ‘others’. Adapted from
Tanji et al. (2007)

13
1100 Brain Structure and Function (2023) 228:1095–1105

Selective advantage tested on a task such as the Raven’s Matrices (Crone et al.
2009). The reason why the activation is in the mid-dor-
Given that the granular PF cortex is unique to primates, it sal PF cortex is that the area is specialized for learning
is important to consider the selection pressures that led to abstract rules concerning sequences (Shima et al. 2007).
its evolution. Primates differ from other mammals in two Sequences are defined by order, and that is why there
critical respects. One is that they have hands with which is activation in the same area when subjects are required
to grasp their food. That is why the mid-dorsal and ventral to re-order or manipulate sequences of letters in their head
PF areas have outputs to the oculomotor system and hand (D'Esposito et al. 1999), calculate using numbers which lie
areas of the premotor system (Borra et al. 2019). on an ordinal scale or perform the n-back task in which the
The second difference is that their food sources exist correct response depends on the order in which the items are
in patches. The caudal PF cortex exists in prosimians as presented (Assem et al. 2020). All these tasks require the
well as monkeys and apes (Wise 2017), and it supports the subjects to appreciate the order in which A is followed by
ability to move the gaze within a patch (Collin et al. 1982). B and C, and 3 by 4 and 5. In other words, the alphabet and
The mid-dorsal and ventral PF cortex are unique to number are arranged in an ordinal sequence.
anthropoid primates (Wise 2017). The mid-dorsal PF cor- An alternative way of testing intelligence is to present
tex supports the ability to prospectively encode an ordered problems that assess semantic knowledge. Wendelken et al.
search using the eye and hand (Owen et al. 1990, 1996; (2008) presented problems in verbal form. For example, the
Passingham 1985). The ventral PF cortex differs in that subjects had to complete ‘brain is to thought as stomach is
it supports the ability to learn which types of fruit and to?’ and when they did so there was activation in the ventral
leaves should be the target for search, these being defined PF cortex. Price and Friston (1999) presented an adapta-
in terms of their look and feel (Passingham 2021). tion of the pyramid and palm trees test (Howard and Patter-
However, during the evolution of the anthropoid pri- son 1992) in which the subject has to decide which of two
mates the availability of particular fruits and leaves varied semantically similar pictures or words is closest to a third
dramatically within a year and from year to year (Chap- semantically related picture or word. The sample and choice
man et al. 1999; Janmaat et al. 2006). This meant that the items were presented simultaneously. The frontal activation
anthropoids faced challenges as the result of the volatility was in the left ventral PF cortex.
of resources. It was these challenges that led to selection However, Price and Friston (1999) argued that this acti-
pressures for the ability to learn abstract rules, and thus the vation was not necessary for correct performance because a
ability to solve novel problems rapidly if they shared an patient with a ventral PF lesion could succeed on the task.
underlying logic with previous problems. To put it another But the tasks were deliberately made easy. That difficulty
way, they promoted intelligence. matters was shown by a study in which rTMS was applied
to the left ventral PF cortex. The subjects were not impaired
if the correct associate ‘salt’ was ‘pepper’; but they were
impaired if the correct associate was ‘grain’ (Whitney et al.
Human intelligence 2010).
These tests of semantic knowledge assess the ability
Human intelligence depends on the fact that our brain has to appreciate associations, and that is why the activations
co-opted this system and has reused it for novel purposes. are in the ventral PF cortex. Just as there is cell activity in
Its potential has been greatly amplified by the fact that this area when macaque monkeys are tested on associative
language provides a tool for instruction and thus formal learning (Asaad et al. 1998; Rainer et al. 1999), so there are
education. The prolonged period of education further activations there when human subjects learn verbal paired
encourages the ability to transfer knowledge from one type associates (Fletcher et al. 1995). Tasks of this sort require
of problem to another. the transformation from A to X and B to Y.
Intelligence can, of course, be assessed by non-verbal
IQ tests. Solving these tests does not necessarily require
facility with language (Woolgar et al. 2018). Indeed a Objections and limitations
patient with severe aphasia and no inner speech could suc-
ceed on the Raven’s Matrices (Levine et al. 1982). It is important to consider possible objections to the
Typically tests of fluid intelligence present sequences approach recommended here. There are three.
of letters, numbers or designs and the task is to find the First, it could be argued that if area A performs a
abstract rule that the sequences obey. There is extensive unique transformation from its inputs to outputs, then
activation in the mid-dorsal PF cortex when subjects are areas X, Y and Z should not be able to take over after
a lesion in area A. Yet, we know that patients with

13
Brain Structure and Function (2023) 228:1095–1105 1101

neurological disorders can show recovery (Sampaio- Conclusions

Baptista et al. 2018; Tomassini et al. 2012). There could
be several explanations. First, if the lesion is unilateral, It is not possible to say whether we understand the PF cor-
the other hemisphere may be able to support recovery. tex in a short paper. It demands a book (Passingham 2021).
For example, monkeys with unilateral mid-dorsal PF But the hope is that reframing the problem as describ-
lesions can relearn delayed responses tasks (Ainsworth ing the transformations that are performed by each region
et al. 2018; Rosen et al. 1975); but monkeys with bilat- within the PF cortex, given specific inputs and outputs,
eral lesions fail to relearn delayed alternation in 2000 helps to clarify what we need to know.
trials (Goldman et al. 1971). Second, recovery is possible So does this way of framing the problem help us to
by learning a new strategy, for example attending to a appreciate why so many apparently different suggestions
location during the delay period rather than depending have been made concerning the functions of the PF cortex?
on short-term memory (Ainsworth et al. 2018). Finally, First, it abolishes the riddle. Then it encompasses the pro-
given the astonishing complexity of the neural networks, posal that the PF cortex supports conditional learning in
alternative routes from vision to action may remain, even which the response depends on a single event (Passingham
if performance is sub-optimal. and Wise 2012). In a sequence A leads to B, whereas in an
Second, it could be objected that if each area performs association A is linked with B. In both cases B is condi-
a transformation from its inputs to its outputs, this is tional on A. Furthermore, in both cases the selection of B
incompatible with the flexibility shown by the anthro- depends on A. And attentional selection via enhancement
poid brain. But that flexibility can be accounted for in of the sensory inputs or specific outputs is an executive
two ways. First, the PF cortex enhances activity in differ- mechanism involved in cognitive control.
ent sensory areas depending on the current task (Heinen However, it is not always possible in nature to select
et al. 2014). Second, the PF cortex enhances activity B immediately. This means that the transformation is
in different output areas depending on the current task delayed, and the representation of B must be encoded and
(Sakai and Passingham 2006). This means that each PF maintained during the delay period. Prospective activity
area has different inputs and outputs as a function of the of this sort was illustrated in Fig. 3, an aspect of working
task demands. memory. Furthermore, A may lead not just to B but also
The third potential objection is that the clearest evi- to C and D, in other words a sequence. And Fig. 4 illus-
dence for distinguishing the mid-dorsal and ventral PF trates activity encoding the upcoming sequence. To put it
cortex comes from lesion studies (Passingham 1975). But another way the activity encodes the plan.
in the intact brain, cell recordings do not indicate such Finally, claims that the PF cortex serves as a multiple-
a clear distinction between areas (Brincat et al. 2018). demand system was originally based on an imaging study
For example, as already mentioned there are cells in the (Fedorenko et al. 2013) in which the statistical analysis
mid-dorsal PF cortex, and not just the ventral PF cortex, was not corrected for multiple comparisons. When Duncan
that encode abstract visual matching rules (Wallis et al. used corrected thresholds (Assem et al. 2020), the activa-
2001). Yet the inferotemporal cortex (Webster et al. 1994) tion was much more restricted. It lay in area 9/46v and
and perirhinal cortex (Kondo et al. 2005) project to the area 46. The activation in the mid-dorsal PF cortex reflects
ventral and orbital PF cortex, but not to the mid-dorsal the fact that two of the three tasks involved sequences.
PF cortex. The implication is that the cell activity in the The activation in area 9/46v reflects the fact that the third
mid-dorsal PF cortex during visual matching is derived task was associative, and there are strong interconnections
from the interconnections between the ventral and mid- between ventral PF cortex and area 9/46v (Petrides and
dorsal PF cortex (Petrides and Pandya 2002). A lesion in Pandya 2002); in other words, the activation is derived.
the ventral PF cortex cuts these connections. We take the description in terms of information flow to
However, this does not mean that in the intact brain be more fundamental than the description using psycho-
the mid-dorsal PF cortex makes no contribution to per- logical terminology because it is not restricted simply to
formance on visual matching. For example, it provides labelling. Instead, it provides an account of how the differ-
one route, though not the only one, to the premotor areas ent PF areas work and how they interact with each other,
(Borra et al. 2019; Takahara et al. 2012). Thus, even given known anatomical inputs and outputs. Future studies
though a lesion that includes the mid-dorsal PF cortex can aim to explain other psychological phenomena within
has no effect on visual matching (Passingham 1975), it the same framework. For example, is the activation in the
still makes a contribution to the intact brain. And it is the ventral PF cortex when subjects are perceptually aware
intact brain we are trying to understand. (Lau and Passingham 2006) to be explained because the
experienced percept (the input) is associated with and thus

13
1102 Brain Structure and Function (2023) 228:1095–1105

codes prospectively for stimuli that can be classed as being References

similar (the output)?
But there are two more requirements for a full under- Ainsworth M et al (2018) Functional reorganisation and recovery
standing. First, we need an account of how each PF area following cortical lesions: a preliminary study in macaque
monkeys. Neuropsychologia 119:382–391
interacts with the rest of the cortical network. We have a Amiez C, Petrides M (2007) Selective involvement of the mid-dor-
very detailed connectome (Ercsey-Ravasz et al. 2013), and solateral prefrontal cortex in the coding of the serial order of
as already mentioned very heavy interconnections have visual stimuli in working memory. Proc Natl Acad Sci USA
been described between the different PF areas (Harriger 104:13786–13791
Andreau JM, Funahashi S (2011) Primate prefrontal neurons encode
et al. 2012). The PF cortex is unique in the number of dif- the association of paired visual stimuli during the pair-associ-
ferent cells that have conjunctive properties (Passingham ation task. Brain Cogn 76:58–69
and Wise 2012), for example, coding both for location and Asaad WF, Rainer G, Miller EK (1998) Neural activity in the pri-
object identity or the stimulus and goal. mate prefrontal cortex during associative learning. Neuron
21:1399–1407
The mechanism for interaction may depend on oscil- Assem M et al (2020) A domain-general cognitive core defined
latory activity. This is ubiquitous throughout the hierar- in multimodally parcellated human cortex. Cereb Cortex
chy, with higher frequencies in the PF cortex (Lundqvist 30:4361–4380
et al. 2020), and it may play a role in changes in func- Averbeck BB, Sohn JW, Lee D (2006) Activity in prefrontal cortex
during dynamic selection of action sequences. Nat Neurosci
tional connectivity between areas depending on the current 9:276–282
task (Pariz et al. 2021). The inputs and outputs are fixed, Baddeley A (1996) Exploring the central executive. Q J Exp Psychol
though they may differ in strength with experience. We 49:5–28
need to know whether the flexibility supported by the PF Blacker KJ, Courtney SM (2016) Distinct neural substrates for main-
taining locations and spatial relations in working memory.
cortex depends on changes in functional connectivity. Front Hum Neurosci 10:594. https://​doi.​org/​10.​3389/​fnhum.​
Second, a full understanding requires an account of how 2016.​00594.​eColl​ectio​n2016
the particular cells within each area interact so as to enable Boettiger CA, D’Esposito M (2005) Frontal networks for learning
its transformation from inputs to outputs. And that account and executing arbitrary stimulus-response associations. J Neu-
rosci 25:2723–2732
may well differ from area to area. Heinzle et al. (2007) Borra E et al (2014) Projections to the superior colliculus from infe-
have produced such an account for the frontal eye field, rior parietal, ventral premotor, and ventrolateral prefrontal
and they specifically compared the columnar structure areas involved in controlling goal-directed hand actions in the
with that in visual areas. Similar detailed comparisons of macaque. Cereb Cortex 24:1054–1065
Borra E et al (2019) Rostro-caudal connectional heterogeneity of the
the columnar structure need to be made in each of the PF dorsal part of the macaque prefrontal area 46. Cereb Cortex
areas and the results incorporated into computational mod- 29:485–504
els of how they perform their particular transformation. Botvinick M et al (2019) Reinforcement learning, fast and slow.
These computational theories need to be tested against Trends Cogn Sci 23:408–422
Brincat SL et al (2018) Gradual progression from sensory to task-
each other to see which makes the best predictions. It is related processing in cerebral cortex. Proc Natl Acad Sci USA
one thing to produce a computational account, but quite 115:E7202–E7211
another to demonstrate that it makes correct and novel Broadbent D (1958) Perception and communication. Oxford Uni-
predictions that other accounts do not. We are still a long versity Press, Oxford
Buckley MJ, Gaffan D (1998) Perirhinal cortex ablation impairs
way from the sorts of neuronal accounts with detailed pre- visual object identification. J Neurosci 18:2268–2275
dictions that Marr (1969) gave us for the cerebellum. And Bussey T, Wise S, Murray E (2001) The role of ventral and orbital
that was over 50 years ago. prefrontal cortex in conditional visuomotor learning and strat-
egy use in rhesus monkeys (Macaca mulatta). Behav Neurosci
115:971–982
Author contributions REP wrote the main manuscript upon HL's initial Cadieu C et al (2007) A model of V4 shape selectivity and invari-
suggestion. HL suggested further edits. ance. J Neurophysiol 98:1733–1750
Chafee MV, Goldman-Rakic PS (1998) Matching patterns of activity
Funding The authors have not disclosed any funding. in primate prefrontal area 8a and parietal area 7ip neurons during
a spatial working memory task. J Neurophysiol 79:2919–2940
Data availability Enquiries about data availability should be directed Chapman CA et al (1999) Fruit and flower phenology at two sites in
to the authors. Kibale National Park, Uganda. J Trop Ecol 15:189–211
Colby CL, Duhamel JR, Goldberg ME (1996) Visual, presaccadic,
and cognitive activation of single neurons in monkey lateral
Declarations intraparietal area. J Neurophysiol 76:2841–2852
Collin NG et al (1982) The role of frontal eye-fields and superior col-
Conflict of interest The authors have not disclosed any conflict of in- liculi in visual search and non-visual search in rhesus monkeys.
terests. Behav Brain Res 4:177–193
Constantinidis C, Qi XL (2018) Representation of spatial and fea-
ture information in the monkey dorsal and ventral prefrontal

13
Brain Structure and Function (2023) 228:1095–1105 1103

cortex. Front Integr Neurosci 12(31):2018. https://​doi.​org/​10.​ association. Front Neural Circuits 7(158):2013. https://​d oi.​
3389/​fnint.​2018.​00031.​eColl​ection org/​10.​3389/​fncir.​2013.​00158.​eColl​ection
Costa VD et al (2015) Reversal learning and dopamine: a bayesian Howard D, Patterson K (1992) Pyramids and Palm trees: a test of
perspective. J Neurosci 35:2407–2416 semantic access for pictures and words. Thames Valley Test
Cromer JA, Roy JE, Miller EK (2010) Representation of multiple, Company, Bury St Edmunds, UK
independent categories in the primate prefrontal cortex. Neu- Hwang J, Romanski LM (2015) Prefrontal neuronal responses during
ron 66:796–807 audiovisual mnemonic processing. J Neurosci 35:960–971
Crone EA et al (2009) Neurocognitive development of relational Janmaat KR, Byrne RW, Zuberbuhler K (2006) Primates take weather
reasoning. Dev Sci 12:55–66 into account when searching for fruits. Curr Biol 16:1232–1237
D’Esposito M et al (1999) Maintenance versus manipulation of infor- Jerde TA et al (2012) Prioritized maps of space in human frontoparietal
mation held in working memory: an event-related fMRI study. cortex. J Neurosci 32:17382–17390
Brain Cogn 41:66–86 Keller EL et al (2008) Effect of inactivation of the cortical frontal eye
Duncan J (2001) An adaptive coding model of neural function in field on saccades generated in a choice response paradigm. J
prefrontal cortex. Nat Rev Neurosci 2:820–829 Neurophysiol 100:2726–2737
Duncan J (2010) The multiple-demand (MD) system of the primate Kondo H, Saleem KS, Price JL (2005) Differential connections of
brain: mental programs for intelligent behaviour. Trends Cogn the perirhinal and parahippocampal cortex with the orbital and
Sci 14:172–179 medial prefrontal networks in macaque monkeys. J Comp Neurol
Ercsey-Ravasz M et al (2013) A predictive network model of cer- 493:479–509
ebral cortical connectivity based on a distance rule. Neuron Kunzle H, Akert K, Wurtz RH (1976) Projection of area 8 (frontal eye
80:184–197 field) to superior colliculus in the monkey. An autoradiographic
Fedorenko E, Duncan J, Kanwisher N (2013) Broad domain gener- study. Brain Res 117:487–492
ality in focal regions of frontal and parietal cortex. Proc Natl Lau HC, Passingham RE (2006) Relative blindsight in normal observ-
Acad Sci USA 110:16616–16621 ers and the neural correlate of visual consciousness. Proc Natl
Ferreira CT et al (1998) Spatio-temporal working memory and fron- Acad Sci USA 103:18763–18768
tal lesions in man. Cortex 34:83–98 Levine DN, Calvanio R, Popovics A (1982) Language in the absence
Fletcher PC et al (1995) Brain systems for encoding and retrieval of of inner speech. Neuropsychol 20:391–409
auditory-verbal memory. An in vivo study in humans. Brain Lowe KA, Schall JD (2018) Functional categories of visuomotor neu-
118(Pt 2):401–416 rons in macaque frontal eye field. eNeuro. https://​doi.​org/​10.​
Freedman DJ et al (2002) Visual categorization and the primate pre- 1523/​ENEURO.​0131-​18.​2018 (eCollection 2018)
frontal cortex: neurophysiology and behavior. J Neurophysiol Lundqvist M, Herman P, Miller EK (2018) Working memory:
88:929–941 delay activity, yes! persistent activity? Maybe not. J Neurosci
Gee AL et al (2008) Neural enhancement and pre-emptive percep- 38:7013–7019
tion: the genesis of attention and the attentional maintenance Lundqvist M, Bastos AM, Miller EK (2020) Preservation and changes
of the cortical salience map. Perception 37:389–400 in oscillatory dynamics across the cortical hierarchy. J Cogn
Genovesio A, Wise SP (2008) The neurophysiology of abstract Neurosci 32:2024–2035
response strategies. In: Bunge SA, Wallis J (eds) Neurosci- Mackey WE, Curtis CE (2017) Distinct contributions by frontal and
ence of rule-guided behavior. Oxford University Press, Oxford, parietal cortices support working memory. Sci Rep 7:6188.
pp 81–106 https://​doi.​org/​10.​1038/​s41598-​017-​06293-x
Gerbella M et al (2013) Connectional heterogeneity of the ventral Mackey WE et al (2016) Human dorsolateral prefrontal cortex is not
part of the macaque area 46. Cereb Cortex 23:967–987 necessary for spatial working memory. J Neurosci 36:2847–2856
Goldman PS et al (1971) Analysis of the delayed-alternation deficit Markov NT, Kennedy H (2013) The importance of being hierarchical.
produced by dorsolateral prefrontal lesions in the rhesus mon- Curr Opin Neurobiol 23:187–194
key. J Comp Physiol Psychol 77:212–220 Marr D (1969) A theory of cerebellar cortex. J Physiol 202:437–470
Goldman-Rakic PS (1995) Cellular basis of working memory. Neu- Menon V, D’Esposito M (2022) The role of PFC networks in cogni-
ron 14:477–485 tive control and executive function. Neuropsychopharmacology
Goldman-Rakic PS (1996) The prefrontal landscape: implications 47:90–103
of functional architecture for understanding human mentation Miller EK, Lundqvist M, Bastos AM (2018) Working memory 2.0.
and the central executive. Philos Trans R Soc Lond B Biol Sci Neuron 100:463–475
351:1445–1453 Mushiake H et al (2006) Activity in the lateral prefrontal cortex
Harlow HF, Warren JM (1952) Formation and transfer of discrimina- reflects multiple steps of future events in action plans. Neuron
tion learning sets. J Comp Physiol Psychol 45:482–489 50:631–641
Harriger L, van den Heuvel MP, Sporns O (2012) Rich club organi- Neubert FX et al (2014) Comparison of human ventral frontal cortex
zation of macaque cerebral cortex and its role in network com- areas for cognitive control and language with areas in monkey
munication. PLoS ONE 7:e46497. https://​d oi.​o rg/​1 0.​1 371/​ frontal cortex. Neuron 81:700–713
journ​al.​pone.​00464​97 (Epub 2012) Owen AM et al (1990) Planning and spatial working memory following
Hedreen JC, DeLong MR (1991) Organization of striatopallidal, frontal lobe lesions in man. Neuropsychologia 28:1021–1034
striatonigral and nigrostriatal projections in the macaque. J Owen AM et al (1996) Double dissociation of memory and executive
Comp Neurol 304:569–595 functions in working memory tasks following frontal excisions,
Heinen K et al (2014) Direct evidence for attention-dependent influ- temporal lobe excisions or amygdala-hippocampectomy in man.
ences of the frontal eye-fields on feature-responsive visual cor- Brain 119:1597–1615
tex. Cereb Cortex 24:2815–2821 Pariz AO, Van Goethem JE, van den Hauwe L, Dillen C, Verlooy J,
Heinzle J, Hepp K, Martin KA (2007) A microcircuit model of the Cosyns P, De Schepper AM (2021) Transmission delays and fre-
frontal eye fields. J Neurosci 27:9341–9353 quency detuning can regulate information flow between brain
Hoshi E (2013) Cortico-basal ganglia networks subserving goal- regions. PLoS Comput Biol 17:e1008129
directed behavior mediated by conditional visuo-goal Passingham RE (1975) Delayed matching after selective prefrontal
lesions in monkeys. Brain Res 92:89–102

13
1104 Brain Structure and Function (2023) 228:1095–1105

Passingham RE (1985) Memory of monkeys ({IMacaca mulatta}) Sala JB, Rama P, Courtney SM (2003) Functional topography of a
with lesions in prefrontal cortex. Behav Neurosci 99:3–21 distributed neural system for spatial and nonspatial information
Passingham RE (1993) The frontal lobes and voluntary action. maintenance in working memory. Neuropsychologia 41:341–356
Oxford University Press, Oxford Saleem KS, Kondo H, Price JL (2008) Complementary circuits con-
Passingham RE (2021) Understanding the prefrontal cortex: selec- necting the orbital and medial prefrontal networks with the tem-
tive advantage, connectivity, and neural operations. Oxford poral, insular, and opercular cortex in the macaque monkey. J
University Press, Oxford Comp Neurol 506:659–693
Passingham RE, Wise SP (2012) The neurobiology of prefrontal Sallet J, Mars RB, Noonan MP, Neubert FX, Jbabdi S, O’Reilly JX,
cortex. Oxford University Press, Oxford Filippini N, Thomas AG, Rushworth MF (2013) The organiza-
Passingham RE, Stephan KE, Kotter R (2002) The anatomical basis tion of dorsal frontal cortex in humans and macaques. J Neurosci
of functional localization in the cortex. Nat Rev Neurosci 33:12255–12274
3:606–616 Sampaio-Baptista C, Sanders ZB, Johansen-Berg H (2018) Structural
Petrides M (1991) Functional specialization within the dorsolat- plasticity in adulthood with motor learning and stroke rehabilita-
eral frontal cortex for serial order memory. Proc Biol Sci tion. Annu Rev Neurosci 41:25–40
246:299–306 Schall JD (1991) Neuronal activity related to visually guided saccades
Petrides M, Pandya DN (1999) Dorsolateral prefrontal cortex: in the frontal eye fields of rhesus monkeys: comparison with
comparative cytoarchitectonic analysis in the human and the supplementary eye fields. J Neurophysiol 66:559–579
macaque brain and corticocortical connection patterns. Eur J Shallice T (1982) Specific impairments of planning. Philos Trans R
Neurosci 11:1011–1036 Soc Lond Ser B 298:199–209
Petrides M, Pandya DN (2002) Comparative cytoarchitectonic analy- Shima K, Isoda M, Muahiake H, Tanji J (2007) Categorization
sis of the human and macaque ventrolateral prefrontal cortex of behavioural sequences in the prefrontal cortex. Nature
and corticocortical connection pattern in the monkey. Eur J 445:315–318
Neurosci 16:291–310 Shipp S (2005) The importance of being agranular: a comparative
Phillips HN, Cope TE, Hughes LE, Zhang J, Rowe JB (2018) Moni- account of visual and motor cortex. Philos Trans R Soc Lond B
toring the past and choosing the future: the prefrontal cortical Biol Sci 360:797–814
influences on voluntary action. Sci Rep 8:7247. https://​doi.​org/​ Siegel M, Buschman TJ, Miller EK (2015) Cortical information flow
10.​1038/​s41598-​018-​25127-y during flexible sensorimotor decisions. Science 348:1352–1355
Pochon J-B et al (2001) The role of dorsolateral prefrontal cortex in Spreng RN et al (2010) Default network activity, coupled with the
the preparation of forthcoming actions: an fMRI study. Cereb frontoparietal control network, supports goal-directed cognition.
Cortex 11:260–266 Neuroimage 53:303–317
Preuss TM (2007) Primate brain evolution in phylogenetic context. Sreenivasan KK, Curtis CE, D’Esposito M (2014) Revisiting the role of
In: Kaas J, Preuss TM (eds) The evolution of nervous systems, persistent neural activity during working memory. Trends Cogn
vol 4. Elsevier, New York, pp 3–34 Sci 18:82–89
Price JL, Drevets WC (2010) Neurocircuitry of mood disorders. Neu- Stanton GB, Bruce CJ, Goldberg ME (1995) Topography of projections
ropsychopharmacology 35:192–216 to posterior cortical areas from the macaque frontal eye fields. J
Price CJ, Friston KJ (1999) Scanning patients with tasks they can Comp Neurol 353:291–305
perform. Hum Brain Mapp 8:102–108 Takahara D, Inoue K, Hirata Y, Miyachi S, Nambu A, Takada M,
Rainer G, Rao SC, Miller EK (1999) Prospective coding for objects Hoshi E (2012) Multisynaptic projections from the ventrolateral
in primate prefrontal cortex. J Neurosci 19:5493–5505 prefrontal cortex to the dorsal premotor cortex in macaques—
Rodriguez NY, McKim TH, Basu D, Ahuja A, Desrochers TM (2022) anatomical substrate for conditional visuomotor behavior. Eur J
Monkey dorsolateral prefrontal cortex represents abstract vis- Neurosci 36:3365–3375
ual sequences during a no-report task. BioRxiv. https://​doi.​org/​ Takeda K, Funahashi S (2004) Population vector analysis of primate
10.​1101/​2022.​o9.​19.​508576 prefrontal activity during spatial working memory. Cereb Cortex
Rosen J et al (1975) Effects of one-stage and serial ablations of the 14:1328–1339
middle third of sulcus principalis on delayed alternation per- Tanji J, Shima K, Mushiake H (2007) Concept-based behavioral
formance in monkeys. J Comp Physiol Psychol 89:1077–1082 planning and the lateral prefrontal cortex. Trends Cogn Sci
Rowe J et al (2000) Prefrontal cortex: response selection or mainte- 11:528–534
nance within working memory. Science 288:1656–1660 Teuber HL (1964) The riddle of frontal lobe function in man. In: War-
Rowe JB et al (2001) Imaging the mental components of a planning ren JM, Akert K (eds) The frontal granular cortex and behavior.
task. Neuropsychologia 39:315–327 McGraw-Hill, New York
Rowe J et al (2002a) Attention to action: specific modulation of cor- Tomassini V, Jhabdi S, Klein JC, Behrens TE, Matthews PM, Rush-
ticocortical interactions in humans. Neuroimage 17:988–998 worth MF, Johansen-Berg H (2012) Neuroplasticity and func-
Rowe J et al (2002b) Attention to action in Parkinson’s disease: tional recovery in multiple sclerosis. Nat Rev Neurol 8:635–646
impaired effective connectivity among frontal cortical regions. Tsujimoto S, Postle BR (2012) The prefrontal cortex and oculomotor
Brain 125:276–289 delayed response: a reconsideration of the “mnemonic scotoma.”
Rushworth M et al (1997) Ventral prefrontal cortex is not essential J Cogn Neurosci 24:627–635
for working memory. J Neurosci 17:4829–4838 Tsujimoto S, Genovesio A, Wise SP (2008) Transient neuronal corre-
Sakai K, Passingham RE (2006) Prefrontal set activity predicts rule- lations underlying goal selection and maintenance in prefrontal
specific neural processing during subsequent cognitive perfor- cortex. Cereb Cortex 18:2748–2761
mance. J Neurosci 26:1211–1218 Unterrainer JM et al (2005) The influence of sex differences and indi-
Sakamoto K, Mushiake H, Saito N, Aihara K, Yano M, Tanji J (2008) vidual task performance on brain activation during planning.
Discharge synchrony during the transition of behavioral goal Neuroimage 24:586–590
representations encoded by discharge rates of prefrontal neu- Wallis JD, Anderson KC, Miller EK (2001) Single neurons in prefron-
rons. Cereb Cortex 18:2036–2045 tal cortex encode abstract rules. Nature 411:953–956

13
Brain Structure and Function (2023) 228:1095–1105 1105

Wang G, Tanifuji M, Tanaka K (1998) Functional architecture in mon- Whitney C, Kirk M, O’Sullivan J, Lambon Ralph MA, Jefferies E
key inferotemporal cortex revealed by in vivo optical imaging. (2010) The neural organization of semantic control: TMS evi-
Neurosci Res 32:33–46 dence for a distributed network in left inferior frontal and poste-
Wang M, Zhang H, Li BM (2000) Deficit in conditional visuomotor rior middle temporal gyrus. Cereb Cortex 21:1066–1075
learning by local infusion of bicuculline into the ventral prefron- Wise SP (2017) The evolution of the prefrontal cortex in early primates
tal cortex in monkeys. Eur J Neurosci 12:3787–3796 and anthropoids. In: Krubitzer L, Kaas J (eds) The evolution of
Wang JX, Kurth-Nelson Z, Kumaran D, Tirumala D, Soyer H, Leibo nervous systems. Elsevier, New York
JZ, Hassabis D, Botvinik M (2018) Prefrontal cortex as a meta- Woolgar A, Parr A, Cusack R, Thompson R, Nimmo-Smith I, Torralva
reinforcement learning system. Nat Neurosci 21:860–868 T, Roca M, Antoun N, Manes F, Duncan J (2018) The multiple-
Warden MR, Miller EK (2010) Task-dependent changes in short-term demand system but not the language system supports fluid intel-
memory in the prefrontal cortex. J Neurosci 30:15801–15810 ligence. Nat Hum Behav 2:200–204
Warren JM (1974) Possibly unique characteristics of learning by pri-
mates. J Hum Evol 3:445–454 Publisher's Note Springer Nature remains neutral with regard to
Webster MJ, Bachevalier J, Ungerleider LG (1994) Connections of jurisdictional claims in published maps and institutional affiliations.
inferior temporal areas TEO and TE with parietal and frontal
cortex in macaque monkeys. Cereb Cortex 4:471–483 Springer Nature or its licensor (e.g. a society or other partner) holds
Wendelken C et al (2008) “Brain is to thought as stomach is to?”: inves- exclusive rights to this article under a publishing agreement with the
tigating the role of rostrolateral prefrontal cortex in relational author(s) or other rightsholder(s); author self-archiving of the accepted
reasoning. J Cogn Neurosci 20:682–693 manuscript version of this article is solely governed by the terms of
Westendorff S, Kaping D, Everling S, Womelsdorf T (2016) Prefrontal such publishing agreement and applicable law.
and anterior cingulate cortex neurons encode attentional targets
even when they do not apparently bias behavior. J Neurophysiol
116:796–811

13