Representation within PFC

1
Representation within PFC

• PFC role in control
• Representation of task / goal / attentional
  information
• These representations bias processing in
  posterior structures
• Nature of organization of these representations
• By function
• Memory vs. inhibition (e.g., Fuster, Diamond,
  others)
• Manipulation vs. maintenance (e.g., Petrides)
• By content
• What vs. where (e.g.,Wilson et al.)
• What vs. when (e.g., Petrides, Barone Joseph)

2
Organization by Function

• Stroop model Memory Inhibition

3
Stroop Task
Verbal Response
Context (PFC)
colors
red
words
green
red
green
Color Stimuli
Word Stimuli
Stimulus
GREEN
4
Organization by Function

• Stroop model Memory Inhibition
• Maintenance vs. manipulation? (Petrides et al.)
• What is manipulation?
• Confounded by representation of sequential
  information

5
Organization by Content

• Object vs. spatial?

6
Spatial vs. Object
Pathways subserving processing of spatial vs.
object information
Delay period activity of units in ventral and
dorsal regions of PFC
7
Organization by Content

• Object vs. spatial?
• Rao et al Rainer et al., etc.
• Issues
• RF size (foveal vs. peripheral)
• Confounds in experimental design (e.g., learning)
• Discrete vs. continuum

8
Organization by Content

• Object vs. spatial?
• Rao et al Rainer et al., etc.
• Issues
• RF size (foveal vs. peripheral)
• Confounds in experimental design (e.g., learning)
• Discrete vs. continuum
• Neuroimaging data?

9
Content-based Dissociations in PFC

• Dissociations reported in PFC based on type of
  information represented in working memory (WM)
• 2 most-frequently proposed WM modality
  dissociations
• Spatial / Non-spatial ltgt Dorsal / Ventral PFC
• Verbal / Non-verbal ltgt Left / Right PFC
• 3 n-back experiments contrasting 3 types of
  stimuli
• each experiment looked for dissociations between
  2 of the types
• exp. 1 Letters vs. Shapes
• exp. 2 Letters vs. Locations
• exp. 3 Shapes vs. Locations

10
Exp. 1 Letters vs. Shapes
s
k
P
S
Time









3 sec ITI
Shapes
Letters
11
Exp. 2 Letters vs.Locations
S
S
r
r
P
P
3-Back location TARGET
s
s
Time
p
p







R
R
4 sec ITI
Locations
Letters
12
Exp. 3 Shapes vs.Locations
Time









3.2 sec ITI
(read words aloud)
Locations
Shapes
13
Meta-Analysis of 3 Experiments

• Results few (if any) convincing dissociations
  been found in any of the 3 experiments between
  any of the stimulus types
• Therefore, pooled all 3 experiments into single
  data set
• co-registered subjects from all experiments
• looked for regions sensitive to WM load within
  each modality
• test for high load (2- or 3-back) gt low load
  (0- or 1-back)
• highlights the overwhelming degree of similarity
  between the areas involved in WM for all of the
  stimulus types, including
• bilateral dorso- and ventro- lateral PFC
• bilateral parietal
• bilateral premotor SMA
• anterior cingulate

14
Working Memory for Letters
15
Working Memory for Shapes
16
Working Memory for Spatial Locations
17
No Modality Dissociations?

• 4 possible hypotheses (or some combination of
  them) could explain failure to detect
  dissociations
• 1) Inadequate spatial resolution?
• NOT LIKELY - dissociations reported elsewhere w/
  PET scans (worse resolution)
• 2) Verbal recoding of non-verbal stimuli?
• MAYBE - but Exp. 3 used articulatory suppression
• 3) Obligatory encoding of multiple dimensions
• 4) Overlap of representations between modalities

18
Organization by Content

• Object vs. spatial?
• Rao et al Rainer et al., etc.
• Issues
• RF size (foveal vs. peripheral)
• Confounds in experimental design (e.g., learning)
• Discrete vs. continuum
• Neuroimaging data?
• Equivocal
• Issues (verbalization, spatial resolution thresh
  olding effects -gt discrete vs. continuum
  revisited
• More complex than we thought
• Abstraction?
• AX-CPT (temporal)
• OReilly model (features vs. categories)

19
Dias et al. (1997)

• ID/ED Task
• Subjects (monkeys) must figure out target
  dimension and feature (WCST-like dynamic
  categorization task).
• Monkeys had either VM or DL PFC lesions, or
  control lesions.
• Dimensions Filled-shape and overlaying line
  shape
• Example Only stimuli with triangle
  filled-shape are targets. Each row represents one
  trial. Boxed stim. is correct response.
• Subj. perform until they reach some criterion
  (90 correct)
• Next Switch target feature in one four ways.

20
Dias, et al. 1997

• New target rule can change either
  INTRADIMENSIONALLY (ID) or EXTRADIMENSIONALLY
  (ED)
• E.g. From one filled-shape to another ID. From
  one filled shape to a line-shape ED
• The change can be a REVERSAL (previously
  unselected feature now becomes selected) or it
  can be a SHIFT (novel feature is now target,
  previous feature becomes unselected).
• N.B. EDS change also involves a reversal of
  dimension.

21
Dias et al., 1997

• RESULTS
• Orbital lesions caused perseverations on IDR
  changes
• DL lesions impaired EDS changes
• No lesions sig. impaired IDS changes
• Interpretation Only IDS change does NOT involve
  any reversal. Thus, PFC must be involved in
  INHIBITION.
• Specifically, VM (orbital) affective assocs.
  Made intradimensionally
• DL attentional shifts to other dimensions.
  Lesions impair ability to inhibit previously
  selected dim. (DLPFC) or feature (orbital PFC).

22
OReilly et al.
23
OReilly et al.
SIMULATION RESULTS
24
Nature of Representations within PFC

• Requirement
• PFC representations must be able to produce a
  wide (if not arbitrarily large) set of
  processing configurations in posterior system,
  that support the flexibility of normal human
  behavior
• Theoretical claim
• Active maintenance imposes constraints that are
  synergistic with the development of such
  representations

25
Effects of Active Maintenance

• ?Independent representations

)
Terminal
?
?
.
?
Television
Synthesizer
Independent (PFC)
Distributed (posterior cortex)
26
Effects of Active Maintenance

• ?Independent representations are combinatorial
  and therefore flexible
• Tension between active maintenance and
  semanticity
• PFC and posterior cortex favor different sides
  of this trade-off
• they work together to exploit the usefulness of
  each