SlideShare a Scribd company logo
1 of 73
Download to read offline
Q1. Does learning symbolic arithmetic inhibit the innate non-
   symbolic approximate abilities? (Sarah)


                                                      No! Quite on the contrary!

The precision of numerical discrimination
(JND or Weber fraction) increases with
age.                                                                            Round numbers accurately
                                                                                discriminated
   2                                                                          1:2




   1

  0.8
                                                                              2:3
  0.6

  0.4
                                                                              3:4
  0.2
                                                                              4:5
   0
                 10          20          30          40          50
                                                                              5:6

        Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007)         0   1   2   3    4      5      6   7   10
        Piazza et al., Cognition 2010; Chinello et al., submitted.                                  Age in years
        Piazza et al., 2004       Pica et al., 2004   Halberda et al., 2008
        Power function fit
Does math
education affect the
      ANS ?
  (disentangling maturation from
        education factors)

The Munduruku is an indigenous
population of the Amazon (Brasil)


- They have number words only up to 5.
- They fail in tasks of exact calculation

- They have a spontaneous capacity for
approximate estimation, comparison,
addition

- As a group, they have a normal non-verbal
magnitude system, even for very large
quantities

                   [Piazza, Pica, Dehaene, in preparation]
36 Munduruku subjects                                                                Performance of Munduruku adults

     [aged from 4 to 67]                                                               Uneducated (n=7)                                   Some education (n=13)
 12 Completely uneducated                                  % larger responses
                                                                    100                                                       100
24 Received some education
                                                                      80                                                      80
                                                                             w = 0.288                                               w = 0.177
                                                                      60                                                      60
                  “choose the larger”
                                                                      40                                                      40
                         *
                                                                      20                                                      20

                                                                        0                                                      0
                                                                            0.7                 1               1.4                 0.7               1             1.4
                                                                                  Ratio of n1 and n2 (log scale)



 Weber fraction                                    Munduruku, uneducated                                Weber fraction
 0.5                                               Munduruku, some education                          0.5
                                                   Italian participants (group means)                                                       r²=26.8%, p=0.001

 0.4                                                                                                  0.4
                                                                                                                                          Math education starts
 0.3                                                                                                  0.3



 0.2                                                                                                  0.2



 0.1                                                                                                  0.1



  0                                                                                                   0.0
       0      10             20         30         40          50            60           70                0         1   2         3       4     5       6     7         8
                                             Age                                                                               Years of Education
• Symbolic and non-symbolic competences go hand in hand during
      development and enhance one another in a form of circular or spiral
      causality




                                                               • More exact
                                    Symbolic
                                    number                     • More linear
Intra Parietal Sulcus               code
neuronal populations
                                    Non-
                                    symbolic
                                    number                       • More approximate
                                    code                         • More compressed
Q2. Can non-symbolic numerical abilities be trained ? Which kinds
of games/manipulations can be used to enhance them? (Timothée)


• We just completed a training study on kindergarteners of 4 to 6 years of age!

• The games was a “matching card game”, whereby children were given a card and
had to match it with the card containing the same number of items among several
distracting cards. It was a small group training – suited for real classroom!

• Results: after a ½ hour group training every week for 4 weeks the acuity of the
approx number system is significantly higher then in a group trained on the same
stimuli but on items’ shapes recognition memory.

• Future project: investigate the impact on learning symbolic numbers (could not
be done because research in Italy and was only funded for one year…thus cannot
do a longitudinal follow-up).

• Starting project ion training Mundurukus involving 4 groups:
1. training with approximate quantities (no symbols)
2. training with exact quantities (no symbols/one-to-one correspondence)
3. training with exact verbal symbols (verbal sequence/verbal counting)
4. training with exact visuo-spatial symbols (abacus-like/visual shape recognition)
Q3. Does making kids aware of their existing abilities help them
feeling learning of symbolic arithmetic less complicated? Role of
meta-cognition and self-esteem in learning (Muriel)

• Having the pupils experiencing that they can COUNT ON THEIR INTUITIONS
should be extremely useful and important to boost their motivation and self-
confidence.


• In domains classically treated as being “hard” such that of mathematics, there
seems to be a strong (and largely unconscious) effect of STEREOTYPE
THREAT. (Italian study of north vs. south stereotype in math abilities)

• Math performance heavily influenced by gender stereotypes (i.e., you do not
even need to know it, your teacher “shows it” … )
Q4. Questioning the current educational system: are we introducing
symbolization and symbolic calculation rules too early? When should
we start teaching symbolic maths? Shall we train the pre-existing
approximate abilities first, or shall we train approximate abilities and
exact calculation at the same time? (Muriel, Marie, Théophile, Asma)

According to the results of the present research, we should propose to:

1. Make children aware that by relying on their intuition of magnitude they can get
   very accurate, even though sometimes only approximate, answers to symbolic
   number problems that may seem very complex (and, in passing, that there are NO
   gender difference in these basic abilities!).

2. Train children in performing approximate calculation

3. Train them to make calculation in the more intuitive way (e.g, subtractions starting
   from the large number and not from the units, using decomposition), and only
   MUCH LATER introducing calculation procedures such that of carrying.

This WILL AVOID the presence of major but frequent calculation ERRORS (e.g., the
result of a subtraction is larger than the subtraend) due to bad understanding of the
calculation procedures, and withdrawing from math BECAUSE of no or little
understanding of calculation procedures.
“Number neurons” in macaque




                              [Nieder, Science 2002]
Recording
neuronal firing   example:
                  Neurons in motor cortex coding the
                  direction of the arm movement




                            Raster-plot
                            Each line corresponds to a trial
                            Each train is an ACTION POTENTIAL (spike)

                            The you calculate the mean across trials
                            (spike rate), and compare spike rates of a
                            given (set of) neuron(s) in different conditions .
                            So you derive responce functions (“tuning
                            curves”).
                            A “tunig curve” for a given stimulus parameter
                            (here movement direction) is a curve
                            describing how the neuron(s) responds to
                            different values of that parameter:




                                   Spikes/sec   0   4   3 2 1 8 7 6 5 4
                                                            directions
Single neurons
recording.in
monkeys
performing the
numerosity
comparison task




                  “Tuning curves” for
                     numerosity
Weber law for
numerosity
coding at the level
of single neuron
tuning curves
Multiple regions contain neurons coding for number. Which does what?




                                                  Responce latency (ms)


 Number is initially extracted from parietal neurons and then the information is
 transmitted to prefrontal cortex neurons.
Key function of PARIETAL CORTEX = DYNAMIC REMAPPING OF SPACE

Pairetal cortex CONTAINS MULTIPLE REPRESENTATIONS OF SPACE
in different egocentric frames of reference FOR ACTION PREPARATION

Spatial location of stimuli are
remapped from the coordinate of the RECEPTOR SURFACE (retina, coclea, ...)
 To the coordinates of the EFFECTOR (eyes, head, hands, ...)




                                                     • Highly plastic (tool use
                                                     changes the receptive field
                                                     of MIP arm-centred
                                                     neurons)

                                                     • Perform operations
                                                     equivalent to vector
                                                     addition
Putative homologies in parietal cortex maps of man and monkeys




Macaque monkey                         VIP (visual-tactile-vestibular-mutlisensory
                                       head centered - NUMEROSITY)
                                        LIP (visual - saccades – eye centered)

                                         AIP (motor-tactile- grasping- hand centered)




Human                                  VIP (multisensory – face – NUMEROSITY?)


                                           LIP (saccades - eyes)

                                        AIP (grasping, hand)
How to study the “neural code” in humans?
                     Principles of fMRI                                                  Control condition
                     (funtional magnetic resonance)
-NUCLEAR MAGNETIC RESONANCE consists in the absorbtion by protons of
Idrogen of electromagnetic waves of given frequency (MegaHz ), in the presence
of a magnetic field. Protons’s spins are usually randomly distributed, while in the
absence of a strong magnetic field align to the directions generated by the
electromagnetic field.

- If we give an electromagnetic impulse at an adequate frequency (dipendent upon
the magnetic field) spin change their rotatio axes. Then they go back to their initial
state. The retourn to the initial equilibrium generates the emission of
electromagnetic waves measurable at distance, which constants of relaxations
(T1, T2) are dependent upon the tissue in which the atom is embedded into.

How to make the RMN signal sensitive to the CEREBRAL ACTIVITY?                           Activity condition

- Deoxi-emoglobin is paramagnetic thus perturbs the RMN signal (effect on T2
apparent, o T2*)

- Brain activity generates:
- Increased oxigen consumption and increased blood supply.
- Oxi/deoxi emoglobine ratio increase
- Magnetic susceptibility decreases
- T2* parameter increases
- RM signal increases


↑Neural activity  ↑ blood flux  ↑ oxi-hemoglobin  ↑ T2*  ↑ BOLD signal
Since BOLD (blood oxygen level dependent) signal is linked to changes in blood
   flow BOLD response is:

1. SLOW compared to the neural response
2. DELAYED compared to the neural response

                         BOLD




                                           seconds
                         stimulus
This link is studied by   Still quite *&^%$#@ clueless   This link is studied by MR
 neurophysiology and is                   here!            physics and approximately
approximately understood                                           understood
BUT….. LUCKILY …
Simultaneous measures of electric NEURAL and fMRI BOLD signals demonstrate
                that the two ARE HIHGLY CORRELATED!!!!!!!




Example: BOLD variation with
stimolus intesity




                                             STRONG CORRELATION
                                             NETWEEN BOLD and
                                             elettrophysiological measures
                                             (1. average on action potentials
                                             over multiple neurons (MUA), and
                                             2. Local field potential (LFP) on
                                             under threashold activity).
Using “adaptation” we can increase spatial
                   resolution
       sampled volume (voxel, typically 2X2X2 mm)
                                                                 tuning curves




                                                                stimulus space



        CLASSIC SUBTRACTION METHOD                                          ADAPTATION METHOD

stimulus S 1                           stimulus S 2      S2 preceded by S 2          S2 preceded by S 1




    total I(S1 )       =        total I(S2 )

    Different populations code for
                                                         total I(S2, S2 )        <   total I(S1,S 2)
                                                      Measurable difference in activation,
    S1 and S2, but the total
                                                      indicating that S1 and S2 are coded by
    activation is = for S1 and S2
                                                      different neural populations
Using “adaptation” we can decipher neural coding
           schemes (“tuning curves”)


                      Adattamento dell’attività neurale

    Firing rate   1       2    3   4   5 6 7 8 9…




        0
Using “adaptation” we can decipher neural coding
           schemes (“tuning curves”)


                      Adattamento dell’attività neurale

    Firing rate   1       2    3   4   5 6 7 8 9…




        0
Using “adaptation” we can decipher neural coding
           schemes (“tuning curves”)


                      Adattamento dell’attività neurale

    Firing rate   1       2    3   4   5 6 7 8 9…




        0
Using “adaptation” we can decipher neural coding
           schemes (“tuning curves”)


                      Adattamento dell’attività neurale

    Firing rate   1       2    3   4   5 6 7 8 9…




        0
                  1       2    3 4 5 6 7 8 9… Test numbers
Using “adaptation” we can decipher neural coding
           schemes (“tuning curves”)


                      Adattamento dell’attività neurale

    Firing rate   1       2    3   4   5 6 7 8 9…




        0
                  1       2    3 4 5 6 7 8 9… Test numbers
Adattamento dell’attività neurale
Corse weberian coding
       Firing rate   1   2   3   4   5 6 7 8 9…




           0
                     1   2   3 4 5 6 7 8 9… Log (test numbers)
Precise weberian coding
      Firing rate    1   2   3   4   5 6 7 8 9…




          0
                     1   2   3 4 5 6 7 8 9… Log (test numbers)
fMRI “adaptation” experiment to investigate
        numerosity coding scheme

                             Adaptation stimuli (16 dots)
                         [150 ms]




                                        [1050 ms]




                                    Deviant stimuli (10% trials)


     8 (far)   10 (medium)          13 (close)      16 (same)       20 (close)        24 (medium)          32 (far)




                                                          Piazza, M. Izard, V., Pinel, P., Le Bihan, D. & Dehaene, S. (2004)
                                                          Neuron
Risponse to deviant numerosities in the IPS
                       bilaterally




 Regions where
activity increases
with a CHANGE
  in NUMBER




                     L            R
Weber law in       Left intraparietal cortex                          Right intraparietal cortex


intraparietal    0.4                                               0.4




sulcus
                 0.2                                               0.2


                   0                                                 0


                -0.2                                 Nadapt 16     -0.2                             Nadapt 16
                                                     Nadaptt 32                                     Nadapt 32
 z = 42
                -0.4                                               -0.4
                       8     16        32                64               8     16    32               64
                 Deviant numerosity (linear scale)                   Deviant numerosity (linear scale)



                 0.4                                               0.4


                 0.2                                               0.2


                   0                                                 0


                -0.2                                   Nadapt 16   -0.2                             Nadapt 16
                                                       Nadapt 32                                    Nadapt 32
                -0.4                                               -0.4
                       8          16            32         64             8          16        32        64
                  Deviant numerosity (log scale)                     Deviant numerosity (log scale)



                0.4                                                0.4


                0.2                                                0.2


                  0                                                  0


                -0.2                                               -0.2


                -0.4                                               -0.4
                       0.5                  1                 2           0.5              1                2
                       Deviation ratio (log scale)                        Deviation ratio (log scale)
Multiple replications using the same paradigm (e.g., Cantlon et al., 2005)




      ADULTS

                                                4 YEARS OLD KIDS




                                       Especially in the RIGHT HIPS!
Risposta alla numerosità nel cervello di
    bebè già a 3 mesi !!! Tecnica dell’EEG




                              A. Experimental design

…




                                                       …




                 Possible test stimuli:
Response to number change in 3 months old babies!! EEG (ERPs)




                                                                RIGHT
Stesso numero
Diversa forma
                Diverso numero
                Stessa forma
                                 Stesso numero Stesso numero
                                 Diversa forma Stessa forma
                                                                HEMISPHE
• WHY IS THIS INTERESTING ? ? ? ? ?
  ???????????????????????????????
• Hp: the non-verbal intuitions of
  NUMEROSITY GROUND our capacity to
  understand numbers and arithmetic
  (Butterworth, Dehaene, etc...)

 If we better understand the cognitive and neural basis
  underlying such start-up-tool we can better understand the
  development of numerical abilities and maybe help
  developing tools which improve teaching efficacy and
  therapeutic tools in cases of dysfunctioning systems (sia
  dello sviluppo che acquisite)
• Criteria for a start-up function / brain
  region:


  (1)-> its integrity should be a necessary condition for
    normal development of symbolic number skills.

  (2)-> its computational constraints should predict the
    speed and ease of symbolic number acquisition.

  (3)-> some traces of its computational signatures may
    be present when humans process symbolic numbers.
• If the ANS grounds the cultural acquisition
  of symbolic number skills it should guide
  and constrain it:

  (1)-> its integrity should be a necessary condition for
    normal development of symbolic number skills.

  (2)-> its computational constraints should predict the
    speed and ease of symbolic number acquisition.

  (3)-> some traces of its computational signatures may
    be present when humans process symbolic numbers.
(1) Traces of the ANS in symbolic
 number processing - behavioural




                          Numbers are treated as
                        representing APPROXIMATE
                        QUANTITIES since the initial
                             stages of learing



                                  Gilmore et al., Nature 2007
(1) Traces of the ANS in symbolic
 number processing - behavioural
Same Ratio-dependent responses
       in non-symbolic and symbolic number
                   processing

                                    ADULTS
                    “choose the larger”               “choose the larger”

                           *                          12     *      16
                                                                               Numbers are treated as
               1               Symbolic comparison                            analogical APPROXIMATE
                                                                            QUANTITIES throughout the life-
             0,95
                                                                                         span
  Accuracy




              0,9
                                          Non-symbolic comparison
             0,85


              0,8


             0,75
                        11.1                2   1.3           3   1.6
                         Ratio (bigger/smaller set)

                        [Chinello et al., under revision]
AFFERMAZIONE:


EVIDENCES (behavioral):

1) “EFFETTO DISTANZA” CON NUMERI
SIMBOLICI


                                                        Tempi di risposta
                      63
                 76
            25
       32

Ai soggetti viene presentato un numero e viene
chiesto di rispondere se sia più grande o più piccolo
di un numero di riferimento (ad es. 65).


  Più piccolo         Più grande

                                                           Errori
 I tempi di risposta e gli errori sono modulati
 dalla distanza (numerica) tra i numeri e
 questo è indice che vi sono tracce di una
 rappresentazione ANALOGICA dei numeri                              Numeri presentati
(1) ANS correlates with symbolic
number processing throughout life-span

                                 Number                               kindergarteners (3 to 6 yoa, N=
                   Finger gnosis Comparison    Visuo-spatial memory
                                                                      94) and of adults (N = 36)
          Grasping
                                  23                                  5 “dorsal” tasks:
                                                                      • visuo-spatial memory (Corsi)
                                                                      • numerosity comparison
                                                                      • symbolic number comparison
                                                                      • finger gnosis
                                                                      • grasping

                                                                      2 “ventral” tasks (Golara et al., 2007):
                                                                      • face recognition memory
                                                                      • object recognition memory
                                                Objects

 [Simon et al., Neuron 2002]           Faces

                                                                                       [Chinello et al., under revision]
Numerosity comparison                                     Finger gnosis                            Face recognition
                                                                                                       4
 2                                                 100
                                                                                                       3
                                                      80
1,5                           R² = 0,26 p<.00                                                          2
                                                      60                            R² = 0,42 p<.00    1




                                                                                                      d'
                                                Error (%)
 1
W




                                                      40                                               0

0,5                                                                                                   -1 3     4            5             6
                                                      20
                                                                                                      -2                R² = 0,07 p<.01
 0                                                          0                                         -3
      3          4            5
                     Age (years)          6                     3      4            5           6                  Age (years)
                                                                           Age (years)
(1) ANS correlates with symbolic
number processing throughout life-span
                                                 …Correlates with math scores up
                                                 to 10 years earlier ...




                                      at 8 yoa
                                                                   at 14 yoa




w measured at 14 years of age … … …
Symbolic number cognition is associated to
               parietal cortex
                                                 Missing gray matter
                                         in premature children with dyscalculia
PARIETAL                                          [Isaacs et al., Brain, 2001]

DYSFUNCTIONS
CAUSE ACALCULIA
                         developmental
 acquired


 Classical lesion site for
        acalculia                          Abnormal gyrification and activation
 [Dehaene et al., TICS, 1997]             in Turner’s syndrome with dyscalculia
                                                   [Molko et al., Neuron, 2003]
Parietal cortex in symbolic number cognition
              PARIETAL ACTIVATION IS SYSTEMATICALLY
                 OBSERVED IN SYMBOLIC NUMBER
                           PROCESSING
        x = - 48                  L     z = 44    z = 49     x = 39                       R        50 %



HIPS



                                                                                                   22 %


• All numerical tasks activate this region
       (e.g. addition, subtraction, comparison, approximation, digit detection…)

• This region fulfils two criteria for a semantic-level representation:
- Format invariance
- Quantity-related

 Crucial parameter coded: numerical quantity
                                                                          [Dehaene, Piazza, Pinel, & Cohen,
                                                                          Cognitive Neuropsychology 2003]
Example of parietal activation “specific” to numbers
(Eger et al, Neuron 2003)
                                    Numbers-(letters&colors)
• Subjects are asked to respond
to a given infrequent stimulus
(number « 5 », letter « B», color
« red »)
• Numbers, letter, and colours
are presented visually and
auditory
•Only non-target stimuli are
analysed
(1) Convergence towards a
  quantity code in the IPS in adults
                     FORMAT                                   NUMBER

                      Deviant format                               Deviant number
                                               Adaptation number
 Adaptation format    dots           digits                         20          50


    Dots              same       = different       17, 18, o 19    close   <    far


    Arabic digits    different   =   same          47, 48, o 49     far    >   close




                      2 CRITERA DEFINITIONAL
                      For a SEMANTIC representation:
                       •INVARIANCE TO ENTRY FORMAT
                      •SEMANTIC METRIC
(1) Convergence towards a
          quantity code in the IPS in adults
                                                                                    1010
                                                                                                 close
                                                                                               close
                                                                                                         Left Parietal Peak
                                                                                                       Right Parietal Peak
Number adaptation protocol                                                                       far
                                                                                           8 8 far
(brain response to a change in number)   DEVIANTS
                                                                                           66

         HABITUATION                        20




                                                               Activation (betas)
                                                                      Activation (betas)
                                                                                           4
                                                                                               4
                                                                                           2
                  18       19                                                                  2
        19
                                                                                           0

                      or                                                             -2
                                                                                               0

                                           50                                                  -2
                                                                                     -4

                                                                                     -6 -4
                                                                                                    DOTS DOTS ARABICARABIC
                                                                                     -8 -6 (among(among (among (among
                                                                                                   dots) arabic) arabic) dots)
                                                                                                   DOTS DOTS ARABIC ARABIC
                                                    Symbolic                                   -8 (among (among (among (among
                                                    number
                                                    code                                              samedifferent
                                                                                                   dots) arabic) arabic) dots)


                                                    Non-
                                                    symbolic
                                                    number
                                                    code




[Piazza et al., Neuron 2007]
• If the ANS grounds the cultural acquisition
  of symbolic number skills it should guide
  and constrain it:

  (1)-> its integrity should be a necessary condition for
    normal development of symbolic number skills.

  (2)-> its computational constraints should predict the
    speed and ease of symbolic number acquisition.

  (3)-> some traces of its computational signatures may
    be present when humans process symbolic numbers.
(2) ANS maturation may account
                                   for lexical acquisition pattern
The precision of numerical discrimination
(JND or Weber fraction) increases with
age.                                                                                                     Round numbers accurately
                                                                                                         discriminated
                            2                                                                          1:2
Estimated weber fraction




                            1

                           0.8
                                                                                                       2:3
                           0.6

                           0.4
                                                                                                       3:4
                           0.2
                                                                                                       4:5
                            0
                                          10          20           30         40          50
                                                                                                       5:6
                                                           Age in years

                                 Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007)         0   1   2   3    4      5      6   7   10
                                 Piazza et al., Cognition 2010; Chinello et al., submitted.                                  Age in years
                                 Piazza et al., 2004       Pica et al., 2004   Halberda et al., 2008
                                 Power function fit
(2) ANS maturation may account
  for lexical acquisition pattern
  In the NUMBER domain, lexical acquisition is a slow and serial process.



  Number words
refer to quantities
               Understand
                   “one” Understand
                           “two” Understand
                                    “three” Understand
        2 years of age                        “four”   Counting principles
                                                         “discovered”
                      3 years of age

                                    4 years of age
Round numbers accurately
1:2       discriminated

                                                                                               Symbolic number
                                                               Number words
                                                                  refer to
                                                                                                 acquisition
                                                                 quantities Understand
                                                                              “one”    Understand
                                                                                         “two” Understand
                                                                                                  “three” Understand
                                                                    2 years of age                          “four”
2:3                                                                                                                  Counting principles
                                                                                                                       “discovered”
                                                                                     3 years of age
3:4
                                                                                                      4 years of age
4:5
5:6

      0    1         2    3     4       5    6    7       10
                              Age in years


          OTS capacity (number of objects attended at a time)
 4
                                                                                         The OTS reaches the adult
 3                                                                                       capacity by 12 months: 4
                                                                                         “attentional pointers”
 2                                                                                       already available. This
                                                                                         does not account for the
 1
                                                                                         lexical acquisition pattern!

               0.5               1               adults
                         Age in years
• If the ANS grounds the cultural acquisition
  of symbolic number skills it should guide
  and constrain it:
  (1)-> its integrity should be a necessary condition for
    normal development of symbolic number skills.

  (2)-> its computational constraints should predict the
    speed and ease of symbolic number acquisition.

  (3)-> some traces of its computational signatures may
    be present when humans process symbolic numbers.
(3)The necessity of ANS for numeracy
      development: dyscalculia
                                                                                                      4 groups of subjects
                                  “choose the larger”
                                                                                 (1) 8-11 years old dyscalculic (diagnosis: Italian standardized
                                               *                                 test), no neurological problems
                                                                                 (2) 8-11 years old matched for IQ and cronological age
                                      n1                n2                       (3) 4-6 years old
                                                                                 (4) Adults


RESULTS (non dyscalculic subjects)

                                      4-6 years                                  8-11 years                               Adults
                          100                                        100                                     100
                                      w=0.34                                 w=0.25                                      w=0.15
                          80                                         80                                      80
% resp « n2 is larger »




                          60                                         60                                      60

                          40                                         40                                      40

                          20                                         20                                      20

                           0                                          0                                       0
                                0.7              1             1.4         0.7           1           1.4           0.7            1            1.4
                                           n1/n2 (log scale)                     n1/n2 (log scale)                         n1/n2 (log scale)

                                                                                                                                          [Piazza et al., Cognition 2010]
(3)The necessity of ANS for numeracy
               development: dyscalculia
                                      “choose the larger”                                                                                                Impairment in the ANS predicts
                                                                                                                                        7
                                                                                                                                                      symbolic number impairement but not
                                              *                                                                                                        performance in other domains (word
                                                                                                                                                                            adults
                                                                                                                                                                            10 yo
                                                                                                                                        6
                                                                                                                                            4
                                                                                                                                                                    reading) yo
                                                                                                                                                                            5




                                                                                                               Distribution Estimates
                                                                                                       N errors in number comparison
                                                                                                                                                                                             10 yo dyscalculics
                                      n1              n2                                                                                5
                                                                                                                                        3,5
                                                                                                                                        4   3
                                      In dyscalculic children the ANS is
                                                                                                                                        2,5
                                      substantially impaired:                                                                           3




                                                                                                                     tasks
                           0,50                                                                                                             2
                                                                                                                                        2
                                                                               non-dyscalculics
                           0,45                                                                                                         1,5
                                                                               dyscalculics
                                                                                                                                        1
                                                                                                                                            1
Estimated weber fraction




                           0,40                                                                                                                                                                   R2 = 0,17
                                                                                                                                        00,5                                                       P=0.04
                                                                                                                                        -0.1      0     0.1   0.2    0.3   0.4   0.5     0.6      0.7   0.8       0.9
                           0,35
                                                                                                                                            0
                                                                                                                                                                     Estimated w
                           0,30                                                                                                                 0,1                 0,3                0,5                    0,7

                           0,25                                                                                                                                       Estimated w
                                                                                      power function
                                                                                        (R2 = 0.97)
                           0,20

                           0,15

                           0,10
                                  0               5         10         15        20               25                   30


                                                                 Age (years)                                                                                               [Piazza et al., Cognition 2010]
Correlations does not imply causation
    The “circular causality” issue
• During development, attaching “meaning” to
  numerical symbols may entail:
  1. Mapping numerical symbols onto pre-existing approximate quantity
  representations.
  2. Refining the quantity representations




• It is thus possible that the core quantity system is:
   –Not only fundational for the acquisition of numerical
   symbols and principles
   –But also deeply modified by the acquisition of
   numerical symbols and numerical principles.
Development of ANS

                            2
                                                                                                       The precision of numerical
                                                                                                       discrimination increases with age.
Estimated weber fraction




                                                                     Power function:                   What is the role of maturation?
                                                                     Exponent = -0.43
                                                                     R2=0.74
                                                                                                       What is the role of education?
                            1
                                                                     p=0.001
                           0.8

                           0.6

                           0.4

                           0.2

                            0
                                       10          20           30          40           50
                                                           Age in years


                                 Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007)
                                 Piazza et al., Cognition 2010; Chinello et al., submitted.
                                 Piazza et al., 2004       Pica et al., 2004  Halberda et al., 2008
                                 Power function fit


                                                                                                             [Piazza & Izard, The Neuroscientist , 2009]
Does math
education affect the
      ANS ?
  (disentangling maturation from
        education factors)

The Munduruku is an indigenous
population of the Amazon (Brasil)


- They have number words only up to 5.
- They fail in tasks of exact calculation

- They have a spontaneous capacity for
approximate estimation, comparison,
addition

- As a group, they have a normal non-verbal
magnitude system, even for very large
quantities

                   [Piazza, Pica, Dehaene, in preparation]
36 Munduruku subjects                                                                Performance of Munduruku adults

     [aged from 4 to 67]                                                               Uneducated (n=7)                                   Some education (n=13)
 12 Completely uneducated                                  % larger responses
                                                                    100                                                       100
24 Received some education
                                                                      80                                                      80
                                                                             w = 0.288                                               w = 0.177
                                                                      60                                                      60
                  “choose the larger”
                                                                      40                                                      40
                         *
                                                                      20                                                      20

                                                                        0                                                      0
                                                                            0.7                 1               1.4                 0.7               1             1.4
                                                                                  Ratio of n1 and n2 (log scale)



 Weber fraction                                    Munduruku, uneducated                                Weber fraction
 0.5                                               Munduruku, some education                          0.5
                                                   Italian participants (group means)                                                       r²=26.8%, p=0.001

 0.4                                                                                                  0.4



 0.3                                                                                                  0.3



 0.2                                                                                                  0.2



 0.1                                                                                                  0.1



  0                                                                                                   0.0
       0      10             20         30         40          50            60           70                0         1   2         3       4     5       6     7         8
                                             Age                                                                               Years of Education
Conclusions
-There is some good evidence for a fundational role of the
parietal system for approximate numerosity in symbolic
numerical representations.

But there is a lot to be discovered:

1) A true causal role of the ANS in dyscalculia awaits confirmation (longitudinal
   studies)

2) What are the neural mechanisms that drive the refinement of the quantity code
   for symbolic stimuli? Are they necessarily mediated by language?

3) Which aspects of maths education enhance approximate number prepresentation
   acuity?
THREE PARIETAL CIRCUITS FOR NUMBER
                      PROCESSING
         Left hemisphere      Axial slice        Right hemisphere
A. x = - 48                z = 44 z = 49    x = 39                  50 %


                                                                           HORIZONTAL SEGMENT OF THE
                                                                           INTRAPARIETAL SULCUS (hips)


                                                                    22 %


B. x = - 49                z = 30           x = 54


                                                                           LEFT ANGUALR GYRUS




C. x = - 26                z = 61           x = 12


                                                                           POSTERIOR SUPERIOR
                                                                           PARIETAL LOBE (more right)
Three parietal circuits for number processing
(Dehaene, Piazza et al.,2003)
                                             CS

      Left hemisphere                                                                   Right hemisphere
                                                         IPS




        HORIZONTAL SEGMENT OF THE
        INTRAPARIETAL SULCUS (HIPS) hVIP?
        -Number comparison
        -Ratio effect                                                   Seen from top
        -Numerical priming
        -Approximate calculation


        LEFT ANGUALR GYRUS (l AG)
        -Retrieval of arithmetical facts (multiplications, additions)
        -Simple exact calculation


        POSTERIOR SUPERIOR
        PARIETAL LOBE (more right)
        (PSPL)  vLIP?
        -Subtractions
        -Complex additions
        -Approximate calculation
Evidence for a verbal code in
     arithmetical facts retrieval
• Interference on TRs in calculation
                             Task1 (arithmetic):
                             Multiplicazions or subtractions
                             Task 2 (short term memory):
                             Phonological (whisper a non-
                             word) o visuo-spatial (remember
                             the position of an object)




                                 Single task
                                 Phonological dual task
                                 Visuo-spatial dual task
Left angular gyrus in arithmetical
               facts retrieval
1. Training experiment: Trained to memorize complex two digits number arithmetical
facts and measure the effects on brain activity




                                                   UNTRAINED
                                                       >
                                                    TRAINED

                                           TRAINED
                                              >
                                          UNTRAINED



                                                                       Ischebeck et al., 2009
Left angular gyrus in arithmetical
               facts retrieval
2. Arithmetical tasks performed in the scanner and activation correlated with
subsequent subjects’ report on the strategy used (fact retreival or computation)




                                                                          [Grabner et al., 2009 ]
Evidence for a spatial code in
    arithmetical computations
• Interference on TRs in calculation
                             Task1 (arithmetic):
                             Multiplicazions or subtractions
                             Task 2 (short term memory):
                             Phonological (whisper a non-
                             word) o visuo-spatial (remember
                             the position of an object)




                                 Single task
                                 Phonological dual task
                                 Visuo-spatial dual task
Evidence for a spatial code in
  arithmetical computations
Do spatial/motor processes interfere with calculation ?




          « Answer the arithmetical problems while
          performing a sequence of finger movements in
          the same time ! »
NO MVTS       MVTS
                                                                     *
                    1200
                                                      *

                    1100



                    1000
        RT (msec)




                     900



                     800



                     700



                     600
                           MULTIPLICATION         ADDITION       SOUSTRACTION




In the dual task, sequential finger movements were found to slow down responses
to additions and subtractions, whereas multiplications (matched for difficulty)
were unaffected
Evidence for a spatial code in
 arithmetical computations: neglect
                                           Modello            Copia del paziente




Regioni corticali tipicamente
dannegggiate nel neglect



                                      Tipico disegno (copia da modello) di un paziente
                                      con negelct




                                  Test della bisezione di linee
Test dello sbarramento di linee
Evidence for a spatial code in
  arithmetical computations: neglect
Numerical bisection test :
“What is the number between 2 and 6?”
“Answer: 5”  RIGHT BIAS!




                                        Zorzi et al., Nature 2002
Posterior parietal
saccade regions in
   calculation
Spatial code in number
representations: the mental number
   line (SNARC effect) is culture
             dependent




                        Shaki et al., 2009 (Psych Bull Rev)

More Related Content

Viewers also liked

Scala for Machine Learning
Scala for Machine LearningScala for Machine Learning
Scala for Machine LearningPatrick Nicolas
 
Machine Learning with Spark MLlib
Machine Learning with Spark MLlibMachine Learning with Spark MLlib
Machine Learning with Spark MLlibTodd McGrath
 
PredictionIO – A Machine Learning Server in Scala – SF Scala
PredictionIO – A Machine Learning Server in Scala – SF ScalaPredictionIO – A Machine Learning Server in Scala – SF Scala
PredictionIO – A Machine Learning Server in Scala – SF Scalapredictionio
 
Machine Learning using Apache Spark MLlib
Machine Learning using Apache Spark MLlibMachine Learning using Apache Spark MLlib
Machine Learning using Apache Spark MLlibIMC Institute
 
MLlib and Machine Learning on Spark
MLlib and Machine Learning on SparkMLlib and Machine Learning on Spark
MLlib and Machine Learning on SparkPetr Zapletal
 
Neural network & its applications
Neural network & its applications Neural network & its applications
Neural network & its applications Ahmed_hashmi
 
Hidden Markov Model & Stock Prediction
Hidden Markov Model & Stock PredictionHidden Markov Model & Stock Prediction
Hidden Markov Model & Stock PredictionDavid Chiu
 
Artificial neural network
Artificial neural networkArtificial neural network
Artificial neural networkDEEPASHRI HK
 

Viewers also liked (11)

Scala for Machine Learning
Scala for Machine LearningScala for Machine Learning
Scala for Machine Learning
 
Python to scala
Python to scalaPython to scala
Python to scala
 
Machine Learning with Spark MLlib
Machine Learning with Spark MLlibMachine Learning with Spark MLlib
Machine Learning with Spark MLlib
 
PredictionIO – A Machine Learning Server in Scala – SF Scala
PredictionIO – A Machine Learning Server in Scala – SF ScalaPredictionIO – A Machine Learning Server in Scala – SF Scala
PredictionIO – A Machine Learning Server in Scala – SF Scala
 
Hidden markov model
Hidden markov modelHidden markov model
Hidden markov model
 
Neural Networks
Neural NetworksNeural Networks
Neural Networks
 
Machine Learning using Apache Spark MLlib
Machine Learning using Apache Spark MLlibMachine Learning using Apache Spark MLlib
Machine Learning using Apache Spark MLlib
 
MLlib and Machine Learning on Spark
MLlib and Machine Learning on SparkMLlib and Machine Learning on Spark
MLlib and Machine Learning on Spark
 
Neural network & its applications
Neural network & its applications Neural network & its applications
Neural network & its applications
 
Hidden Markov Model & Stock Prediction
Hidden Markov Model & Stock PredictionHidden Markov Model & Stock Prediction
Hidden Markov Model & Stock Prediction
 
Artificial neural network
Artificial neural networkArtificial neural network
Artificial neural network
 

Similar to Does math education affect approximate number sense

Shape contexts
Shape contextsShape contexts
Shape contextshuebesao
 
The Origin of Diversity - Thinking with Chaotic Walk
The Origin of Diversity - Thinking with Chaotic WalkThe Origin of Diversity - Thinking with Chaotic Walk
The Origin of Diversity - Thinking with Chaotic WalkTakashi Iba
 
Oak hill presentation
Oak hill presentationOak hill presentation
Oak hill presentationjan4tarheels
 
Why we don’t know how many colors there are
Why we don’t know how many colors there areWhy we don’t know how many colors there are
Why we don’t know how many colors there areJan Morovic
 
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...Martin Rehm
 
Raising Reading Levels: Lessons from a high-poverty high school
Raising Reading Levels: Lessons from a high-poverty high schoolRaising Reading Levels: Lessons from a high-poverty high school
Raising Reading Levels: Lessons from a high-poverty high schoolWilliam S Roulston
 

Similar to Does math education affect approximate number sense (7)

Shape contexts
Shape contextsShape contexts
Shape contexts
 
The Origin of Diversity - Thinking with Chaotic Walk
The Origin of Diversity - Thinking with Chaotic WalkThe Origin of Diversity - Thinking with Chaotic Walk
The Origin of Diversity - Thinking with Chaotic Walk
 
Oak hill presentation
Oak hill presentationOak hill presentation
Oak hill presentation
 
Why we don’t know how many colors there are
Why we don’t know how many colors there areWhy we don’t know how many colors there are
Why we don’t know how many colors there are
 
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...
Comparing Communities of Learning: Incoming Bachelor Students & Working Profe...
 
03.20.2013 - Flavio Cunha
03.20.2013 - Flavio Cunha03.20.2013 - Flavio Cunha
03.20.2013 - Flavio Cunha
 
Raising Reading Levels: Lessons from a high-poverty high school
Raising Reading Levels: Lessons from a high-poverty high schoolRaising Reading Levels: Lessons from a high-poverty high school
Raising Reading Levels: Lessons from a high-poverty high school
 

More from elena.pasquinelli (20)

Gdp2 2013 14-14bis
Gdp2 2013 14-14bisGdp2 2013 14-14bis
Gdp2 2013 14-14bis
 
Gdp2 2013 14-13
Gdp2 2013 14-13Gdp2 2013 14-13
Gdp2 2013 14-13
 
Gdp2 2013 14-14
Gdp2 2013 14-14Gdp2 2013 14-14
Gdp2 2013 14-14
 
Gdp2 2013 14-9
Gdp2 2013 14-9Gdp2 2013 14-9
Gdp2 2013 14-9
 
Gdp2 2013 14-8
Gdp2 2013 14-8Gdp2 2013 14-8
Gdp2 2013 14-8
 
Gdp2 2013 14-7
Gdp2 2013 14-7Gdp2 2013 14-7
Gdp2 2013 14-7
 
Gdp2 2013 14_3
Gdp2 2013 14_3Gdp2 2013 14_3
Gdp2 2013 14_3
 
Gdp2 2013 14-2
Gdp2 2013 14-2Gdp2 2013 14-2
Gdp2 2013 14-2
 
Gdp2 2013 14-1
Gdp2 2013 14-1Gdp2 2013 14-1
Gdp2 2013 14-1
 
20130112
2013011220130112
20130112
 
technomyths
technomythstechnomyths
technomyths
 
Illusions cognitives
Illusions cognitivesIllusions cognitives
Illusions cognitives
 
rabbit
rabbitrabbit
rabbit
 
illusions perceptives
illusions perceptivesillusions perceptives
illusions perceptives
 
Ecc2012 13 10
Ecc2012 13 10Ecc2012 13 10
Ecc2012 13 10
 
ECC_2012 2013_huron
ECC_2012 2013_huronECC_2012 2013_huron
ECC_2012 2013_huron
 
Piazza cogmaster cognitive_neuroscience2013
Piazza cogmaster cognitive_neuroscience2013Piazza cogmaster cognitive_neuroscience2013
Piazza cogmaster cognitive_neuroscience2013
 
20121108 gdp2 fr
20121108 gdp2 fr20121108 gdp2 fr
20121108 gdp2 fr
 
Ecc2012 13 5
Ecc2012 13 5Ecc2012 13 5
Ecc2012 13 5
 
Ecc2012 13 9
Ecc2012 13 9Ecc2012 13 9
Ecc2012 13 9
 

Does math education affect approximate number sense

  • 1. Q1. Does learning symbolic arithmetic inhibit the innate non- symbolic approximate abilities? (Sarah) No! Quite on the contrary! The precision of numerical discrimination (JND or Weber fraction) increases with age. Round numbers accurately discriminated 2 1:2 1 0.8 2:3 0.6 0.4 3:4 0.2 4:5 0 10 20 30 40 50 5:6 Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007) 0 1 2 3 4 5 6 7 10 Piazza et al., Cognition 2010; Chinello et al., submitted. Age in years Piazza et al., 2004 Pica et al., 2004 Halberda et al., 2008 Power function fit
  • 2. Does math education affect the ANS ? (disentangling maturation from education factors) The Munduruku is an indigenous population of the Amazon (Brasil) - They have number words only up to 5. - They fail in tasks of exact calculation - They have a spontaneous capacity for approximate estimation, comparison, addition - As a group, they have a normal non-verbal magnitude system, even for very large quantities [Piazza, Pica, Dehaene, in preparation]
  • 3. 36 Munduruku subjects Performance of Munduruku adults [aged from 4 to 67] Uneducated (n=7) Some education (n=13) 12 Completely uneducated % larger responses 100 100 24 Received some education 80 80 w = 0.288 w = 0.177 60 60 “choose the larger” 40 40 * 20 20 0 0 0.7 1 1.4 0.7 1 1.4 Ratio of n1 and n2 (log scale) Weber fraction Munduruku, uneducated Weber fraction 0.5 Munduruku, some education 0.5 Italian participants (group means) r²=26.8%, p=0.001 0.4 0.4 Math education starts 0.3 0.3 0.2 0.2 0.1 0.1 0 0.0 0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7 8 Age Years of Education
  • 4. • Symbolic and non-symbolic competences go hand in hand during development and enhance one another in a form of circular or spiral causality • More exact Symbolic number • More linear Intra Parietal Sulcus code neuronal populations Non- symbolic number • More approximate code • More compressed
  • 5. Q2. Can non-symbolic numerical abilities be trained ? Which kinds of games/manipulations can be used to enhance them? (Timothée) • We just completed a training study on kindergarteners of 4 to 6 years of age! • The games was a “matching card game”, whereby children were given a card and had to match it with the card containing the same number of items among several distracting cards. It was a small group training – suited for real classroom! • Results: after a ½ hour group training every week for 4 weeks the acuity of the approx number system is significantly higher then in a group trained on the same stimuli but on items’ shapes recognition memory. • Future project: investigate the impact on learning symbolic numbers (could not be done because research in Italy and was only funded for one year…thus cannot do a longitudinal follow-up). • Starting project ion training Mundurukus involving 4 groups: 1. training with approximate quantities (no symbols) 2. training with exact quantities (no symbols/one-to-one correspondence) 3. training with exact verbal symbols (verbal sequence/verbal counting) 4. training with exact visuo-spatial symbols (abacus-like/visual shape recognition)
  • 6. Q3. Does making kids aware of their existing abilities help them feeling learning of symbolic arithmetic less complicated? Role of meta-cognition and self-esteem in learning (Muriel) • Having the pupils experiencing that they can COUNT ON THEIR INTUITIONS should be extremely useful and important to boost their motivation and self- confidence. • In domains classically treated as being “hard” such that of mathematics, there seems to be a strong (and largely unconscious) effect of STEREOTYPE THREAT. (Italian study of north vs. south stereotype in math abilities) • Math performance heavily influenced by gender stereotypes (i.e., you do not even need to know it, your teacher “shows it” … )
  • 7. Q4. Questioning the current educational system: are we introducing symbolization and symbolic calculation rules too early? When should we start teaching symbolic maths? Shall we train the pre-existing approximate abilities first, or shall we train approximate abilities and exact calculation at the same time? (Muriel, Marie, Théophile, Asma) According to the results of the present research, we should propose to: 1. Make children aware that by relying on their intuition of magnitude they can get very accurate, even though sometimes only approximate, answers to symbolic number problems that may seem very complex (and, in passing, that there are NO gender difference in these basic abilities!). 2. Train children in performing approximate calculation 3. Train them to make calculation in the more intuitive way (e.g, subtractions starting from the large number and not from the units, using decomposition), and only MUCH LATER introducing calculation procedures such that of carrying. This WILL AVOID the presence of major but frequent calculation ERRORS (e.g., the result of a subtraction is larger than the subtraend) due to bad understanding of the calculation procedures, and withdrawing from math BECAUSE of no or little understanding of calculation procedures.
  • 8. “Number neurons” in macaque [Nieder, Science 2002]
  • 9. Recording neuronal firing example: Neurons in motor cortex coding the direction of the arm movement Raster-plot Each line corresponds to a trial Each train is an ACTION POTENTIAL (spike) The you calculate the mean across trials (spike rate), and compare spike rates of a given (set of) neuron(s) in different conditions . So you derive responce functions (“tuning curves”). A “tunig curve” for a given stimulus parameter (here movement direction) is a curve describing how the neuron(s) responds to different values of that parameter: Spikes/sec 0 4 3 2 1 8 7 6 5 4 directions
  • 11. Weber law for numerosity coding at the level of single neuron tuning curves
  • 12. Multiple regions contain neurons coding for number. Which does what? Responce latency (ms) Number is initially extracted from parietal neurons and then the information is transmitted to prefrontal cortex neurons.
  • 13. Key function of PARIETAL CORTEX = DYNAMIC REMAPPING OF SPACE Pairetal cortex CONTAINS MULTIPLE REPRESENTATIONS OF SPACE in different egocentric frames of reference FOR ACTION PREPARATION Spatial location of stimuli are remapped from the coordinate of the RECEPTOR SURFACE (retina, coclea, ...)  To the coordinates of the EFFECTOR (eyes, head, hands, ...) • Highly plastic (tool use changes the receptive field of MIP arm-centred neurons) • Perform operations equivalent to vector addition
  • 14. Putative homologies in parietal cortex maps of man and monkeys Macaque monkey VIP (visual-tactile-vestibular-mutlisensory head centered - NUMEROSITY) LIP (visual - saccades – eye centered) AIP (motor-tactile- grasping- hand centered) Human VIP (multisensory – face – NUMEROSITY?) LIP (saccades - eyes) AIP (grasping, hand)
  • 15. How to study the “neural code” in humans? Principles of fMRI Control condition (funtional magnetic resonance) -NUCLEAR MAGNETIC RESONANCE consists in the absorbtion by protons of Idrogen of electromagnetic waves of given frequency (MegaHz ), in the presence of a magnetic field. Protons’s spins are usually randomly distributed, while in the absence of a strong magnetic field align to the directions generated by the electromagnetic field. - If we give an electromagnetic impulse at an adequate frequency (dipendent upon the magnetic field) spin change their rotatio axes. Then they go back to their initial state. The retourn to the initial equilibrium generates the emission of electromagnetic waves measurable at distance, which constants of relaxations (T1, T2) are dependent upon the tissue in which the atom is embedded into. How to make the RMN signal sensitive to the CEREBRAL ACTIVITY? Activity condition - Deoxi-emoglobin is paramagnetic thus perturbs the RMN signal (effect on T2 apparent, o T2*) - Brain activity generates: - Increased oxigen consumption and increased blood supply. - Oxi/deoxi emoglobine ratio increase - Magnetic susceptibility decreases - T2* parameter increases - RM signal increases ↑Neural activity  ↑ blood flux  ↑ oxi-hemoglobin  ↑ T2*  ↑ BOLD signal
  • 16. Since BOLD (blood oxygen level dependent) signal is linked to changes in blood flow BOLD response is: 1. SLOW compared to the neural response 2. DELAYED compared to the neural response BOLD seconds stimulus
  • 17. This link is studied by Still quite *&^%$#@ clueless This link is studied by MR neurophysiology and is here! physics and approximately approximately understood understood
  • 18. BUT….. LUCKILY … Simultaneous measures of electric NEURAL and fMRI BOLD signals demonstrate that the two ARE HIHGLY CORRELATED!!!!!!! Example: BOLD variation with stimolus intesity STRONG CORRELATION NETWEEN BOLD and elettrophysiological measures (1. average on action potentials over multiple neurons (MUA), and 2. Local field potential (LFP) on under threashold activity).
  • 19. Using “adaptation” we can increase spatial resolution sampled volume (voxel, typically 2X2X2 mm) tuning curves stimulus space CLASSIC SUBTRACTION METHOD ADAPTATION METHOD stimulus S 1 stimulus S 2 S2 preceded by S 2 S2 preceded by S 1 total I(S1 ) = total I(S2 ) Different populations code for total I(S2, S2 ) < total I(S1,S 2) Measurable difference in activation, S1 and S2, but the total indicating that S1 and S2 are coded by activation is = for S1 and S2 different neural populations
  • 20. Using “adaptation” we can decipher neural coding schemes (“tuning curves”) Adattamento dell’attività neurale Firing rate 1 2 3 4 5 6 7 8 9… 0
  • 21. Using “adaptation” we can decipher neural coding schemes (“tuning curves”) Adattamento dell’attività neurale Firing rate 1 2 3 4 5 6 7 8 9… 0
  • 22. Using “adaptation” we can decipher neural coding schemes (“tuning curves”) Adattamento dell’attività neurale Firing rate 1 2 3 4 5 6 7 8 9… 0
  • 23. Using “adaptation” we can decipher neural coding schemes (“tuning curves”) Adattamento dell’attività neurale Firing rate 1 2 3 4 5 6 7 8 9… 0 1 2 3 4 5 6 7 8 9… Test numbers
  • 24. Using “adaptation” we can decipher neural coding schemes (“tuning curves”) Adattamento dell’attività neurale Firing rate 1 2 3 4 5 6 7 8 9… 0 1 2 3 4 5 6 7 8 9… Test numbers
  • 25. Adattamento dell’attività neurale Corse weberian coding Firing rate 1 2 3 4 5 6 7 8 9… 0 1 2 3 4 5 6 7 8 9… Log (test numbers) Precise weberian coding Firing rate 1 2 3 4 5 6 7 8 9… 0 1 2 3 4 5 6 7 8 9… Log (test numbers)
  • 26. fMRI “adaptation” experiment to investigate numerosity coding scheme Adaptation stimuli (16 dots) [150 ms] [1050 ms] Deviant stimuli (10% trials) 8 (far) 10 (medium) 13 (close) 16 (same) 20 (close) 24 (medium) 32 (far) Piazza, M. Izard, V., Pinel, P., Le Bihan, D. & Dehaene, S. (2004) Neuron
  • 27. Risponse to deviant numerosities in the IPS bilaterally Regions where activity increases with a CHANGE in NUMBER L R
  • 28. Weber law in Left intraparietal cortex Right intraparietal cortex intraparietal 0.4 0.4 sulcus 0.2 0.2 0 0 -0.2 Nadapt 16 -0.2 Nadapt 16 Nadaptt 32 Nadapt 32 z = 42 -0.4 -0.4 8 16 32 64 8 16 32 64 Deviant numerosity (linear scale) Deviant numerosity (linear scale) 0.4 0.4 0.2 0.2 0 0 -0.2 Nadapt 16 -0.2 Nadapt 16 Nadapt 32 Nadapt 32 -0.4 -0.4 8 16 32 64 8 16 32 64 Deviant numerosity (log scale) Deviant numerosity (log scale) 0.4 0.4 0.2 0.2 0 0 -0.2 -0.2 -0.4 -0.4 0.5 1 2 0.5 1 2 Deviation ratio (log scale) Deviation ratio (log scale)
  • 29.
  • 30.
  • 31. Multiple replications using the same paradigm (e.g., Cantlon et al., 2005) ADULTS 4 YEARS OLD KIDS Especially in the RIGHT HIPS!
  • 32. Risposta alla numerosità nel cervello di bebè già a 3 mesi !!! Tecnica dell’EEG A. Experimental design … … Possible test stimuli:
  • 33. Response to number change in 3 months old babies!! EEG (ERPs) RIGHT Stesso numero Diversa forma Diverso numero Stessa forma Stesso numero Stesso numero Diversa forma Stessa forma HEMISPHE
  • 34. • WHY IS THIS INTERESTING ? ? ? ? ? ???????????????????????????????
  • 35. • Hp: the non-verbal intuitions of NUMEROSITY GROUND our capacity to understand numbers and arithmetic (Butterworth, Dehaene, etc...)  If we better understand the cognitive and neural basis underlying such start-up-tool we can better understand the development of numerical abilities and maybe help developing tools which improve teaching efficacy and therapeutic tools in cases of dysfunctioning systems (sia dello sviluppo che acquisite)
  • 36. • Criteria for a start-up function / brain region: (1)-> its integrity should be a necessary condition for normal development of symbolic number skills. (2)-> its computational constraints should predict the speed and ease of symbolic number acquisition. (3)-> some traces of its computational signatures may be present when humans process symbolic numbers.
  • 37. • If the ANS grounds the cultural acquisition of symbolic number skills it should guide and constrain it: (1)-> its integrity should be a necessary condition for normal development of symbolic number skills. (2)-> its computational constraints should predict the speed and ease of symbolic number acquisition. (3)-> some traces of its computational signatures may be present when humans process symbolic numbers.
  • 38. (1) Traces of the ANS in symbolic number processing - behavioural Numbers are treated as representing APPROXIMATE QUANTITIES since the initial stages of learing Gilmore et al., Nature 2007
  • 39. (1) Traces of the ANS in symbolic number processing - behavioural Same Ratio-dependent responses in non-symbolic and symbolic number processing ADULTS “choose the larger” “choose the larger” * 12 * 16 Numbers are treated as 1 Symbolic comparison analogical APPROXIMATE QUANTITIES throughout the life- 0,95 span Accuracy 0,9 Non-symbolic comparison 0,85 0,8 0,75 11.1 2 1.3 3 1.6 Ratio (bigger/smaller set) [Chinello et al., under revision]
  • 40. AFFERMAZIONE: EVIDENCES (behavioral): 1) “EFFETTO DISTANZA” CON NUMERI SIMBOLICI Tempi di risposta 63 76 25 32 Ai soggetti viene presentato un numero e viene chiesto di rispondere se sia più grande o più piccolo di un numero di riferimento (ad es. 65). Più piccolo Più grande Errori I tempi di risposta e gli errori sono modulati dalla distanza (numerica) tra i numeri e questo è indice che vi sono tracce di una rappresentazione ANALOGICA dei numeri Numeri presentati
  • 41. (1) ANS correlates with symbolic number processing throughout life-span Number kindergarteners (3 to 6 yoa, N= Finger gnosis Comparison Visuo-spatial memory 94) and of adults (N = 36) Grasping 23 5 “dorsal” tasks: • visuo-spatial memory (Corsi) • numerosity comparison • symbolic number comparison • finger gnosis • grasping 2 “ventral” tasks (Golara et al., 2007): • face recognition memory • object recognition memory Objects [Simon et al., Neuron 2002] Faces [Chinello et al., under revision]
  • 42. Numerosity comparison Finger gnosis Face recognition 4 2 100 3 80 1,5 R² = 0,26 p<.00 2 60 R² = 0,42 p<.00 1 d' Error (%) 1 W 40 0 0,5 -1 3 4 5 6 20 -2 R² = 0,07 p<.01 0 0 -3 3 4 5 Age (years) 6 3 4 5 6 Age (years) Age (years)
  • 43. (1) ANS correlates with symbolic number processing throughout life-span …Correlates with math scores up to 10 years earlier ... at 8 yoa at 14 yoa w measured at 14 years of age … … …
  • 44. Symbolic number cognition is associated to parietal cortex Missing gray matter in premature children with dyscalculia PARIETAL [Isaacs et al., Brain, 2001] DYSFUNCTIONS CAUSE ACALCULIA developmental acquired Classical lesion site for acalculia Abnormal gyrification and activation [Dehaene et al., TICS, 1997] in Turner’s syndrome with dyscalculia [Molko et al., Neuron, 2003]
  • 45. Parietal cortex in symbolic number cognition PARIETAL ACTIVATION IS SYSTEMATICALLY OBSERVED IN SYMBOLIC NUMBER PROCESSING x = - 48 L z = 44 z = 49 x = 39 R 50 % HIPS 22 % • All numerical tasks activate this region (e.g. addition, subtraction, comparison, approximation, digit detection…) • This region fulfils two criteria for a semantic-level representation: - Format invariance - Quantity-related  Crucial parameter coded: numerical quantity [Dehaene, Piazza, Pinel, & Cohen, Cognitive Neuropsychology 2003]
  • 46. Example of parietal activation “specific” to numbers (Eger et al, Neuron 2003) Numbers-(letters&colors) • Subjects are asked to respond to a given infrequent stimulus (number « 5 », letter « B», color « red ») • Numbers, letter, and colours are presented visually and auditory •Only non-target stimuli are analysed
  • 47. (1) Convergence towards a quantity code in the IPS in adults FORMAT NUMBER Deviant format Deviant number Adaptation number Adaptation format dots digits 20 50 Dots same = different 17, 18, o 19 close < far Arabic digits different = same 47, 48, o 49 far > close 2 CRITERA DEFINITIONAL For a SEMANTIC representation: •INVARIANCE TO ENTRY FORMAT •SEMANTIC METRIC
  • 48. (1) Convergence towards a quantity code in the IPS in adults 1010 close close Left Parietal Peak Right Parietal Peak Number adaptation protocol far 8 8 far (brain response to a change in number) DEVIANTS 66 HABITUATION 20 Activation (betas) Activation (betas) 4 4 2 18 19 2 19 0 or -2 0 50 -2 -4 -6 -4 DOTS DOTS ARABICARABIC -8 -6 (among(among (among (among dots) arabic) arabic) dots) DOTS DOTS ARABIC ARABIC Symbolic -8 (among (among (among (among number code samedifferent dots) arabic) arabic) dots) Non- symbolic number code [Piazza et al., Neuron 2007]
  • 49. • If the ANS grounds the cultural acquisition of symbolic number skills it should guide and constrain it: (1)-> its integrity should be a necessary condition for normal development of symbolic number skills. (2)-> its computational constraints should predict the speed and ease of symbolic number acquisition. (3)-> some traces of its computational signatures may be present when humans process symbolic numbers.
  • 50. (2) ANS maturation may account for lexical acquisition pattern The precision of numerical discrimination (JND or Weber fraction) increases with age. Round numbers accurately discriminated 2 1:2 Estimated weber fraction 1 0.8 2:3 0.6 0.4 3:4 0.2 4:5 0 10 20 30 40 50 5:6 Age in years Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007) 0 1 2 3 4 5 6 7 10 Piazza et al., Cognition 2010; Chinello et al., submitted. Age in years Piazza et al., 2004 Pica et al., 2004 Halberda et al., 2008 Power function fit
  • 51. (2) ANS maturation may account for lexical acquisition pattern In the NUMBER domain, lexical acquisition is a slow and serial process. Number words refer to quantities Understand “one” Understand “two” Understand “three” Understand 2 years of age “four” Counting principles “discovered” 3 years of age 4 years of age
  • 52. Round numbers accurately 1:2 discriminated Symbolic number Number words refer to acquisition quantities Understand “one” Understand “two” Understand “three” Understand 2 years of age “four” 2:3 Counting principles “discovered” 3 years of age 3:4 4 years of age 4:5 5:6 0 1 2 3 4 5 6 7 10 Age in years OTS capacity (number of objects attended at a time) 4 The OTS reaches the adult 3 capacity by 12 months: 4 “attentional pointers” 2 already available. This does not account for the 1 lexical acquisition pattern! 0.5 1 adults Age in years
  • 53. • If the ANS grounds the cultural acquisition of symbolic number skills it should guide and constrain it: (1)-> its integrity should be a necessary condition for normal development of symbolic number skills. (2)-> its computational constraints should predict the speed and ease of symbolic number acquisition. (3)-> some traces of its computational signatures may be present when humans process symbolic numbers.
  • 54. (3)The necessity of ANS for numeracy development: dyscalculia 4 groups of subjects “choose the larger” (1) 8-11 years old dyscalculic (diagnosis: Italian standardized * test), no neurological problems (2) 8-11 years old matched for IQ and cronological age n1 n2 (3) 4-6 years old (4) Adults RESULTS (non dyscalculic subjects) 4-6 years 8-11 years Adults 100 100 100 w=0.34 w=0.25 w=0.15 80 80 80 % resp « n2 is larger » 60 60 60 40 40 40 20 20 20 0 0 0 0.7 1 1.4 0.7 1 1.4 0.7 1 1.4 n1/n2 (log scale) n1/n2 (log scale) n1/n2 (log scale) [Piazza et al., Cognition 2010]
  • 55. (3)The necessity of ANS for numeracy development: dyscalculia “choose the larger” Impairment in the ANS predicts 7 symbolic number impairement but not * performance in other domains (word adults 10 yo 6 4 reading) yo 5 Distribution Estimates N errors in number comparison 10 yo dyscalculics n1 n2 5 3,5 4 3 In dyscalculic children the ANS is 2,5 substantially impaired: 3 tasks 0,50 2 2 non-dyscalculics 0,45 1,5 dyscalculics 1 1 Estimated weber fraction 0,40 R2 = 0,17 00,5 P=0.04 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0,35 0 Estimated w 0,30 0,1 0,3 0,5 0,7 0,25 Estimated w power function (R2 = 0.97) 0,20 0,15 0,10 0 5 10 15 20 25 30 Age (years) [Piazza et al., Cognition 2010]
  • 56. Correlations does not imply causation The “circular causality” issue • During development, attaching “meaning” to numerical symbols may entail: 1. Mapping numerical symbols onto pre-existing approximate quantity representations. 2. Refining the quantity representations • It is thus possible that the core quantity system is: –Not only fundational for the acquisition of numerical symbols and principles –But also deeply modified by the acquisition of numerical symbols and numerical principles.
  • 57. Development of ANS 2 The precision of numerical discrimination increases with age. Estimated weber fraction Power function: What is the role of maturation? Exponent = -0.43 R2=0.74 What is the role of education? 1 p=0.001 0.8 0.6 0.4 0.2 0 10 20 30 40 50 Age in years Infants (Izard et al., 2009; Xu & Spelke, 2000; Xu & Arriaga, 2007) Piazza et al., Cognition 2010; Chinello et al., submitted. Piazza et al., 2004 Pica et al., 2004 Halberda et al., 2008 Power function fit [Piazza & Izard, The Neuroscientist , 2009]
  • 58. Does math education affect the ANS ? (disentangling maturation from education factors) The Munduruku is an indigenous population of the Amazon (Brasil) - They have number words only up to 5. - They fail in tasks of exact calculation - They have a spontaneous capacity for approximate estimation, comparison, addition - As a group, they have a normal non-verbal magnitude system, even for very large quantities [Piazza, Pica, Dehaene, in preparation]
  • 59. 36 Munduruku subjects Performance of Munduruku adults [aged from 4 to 67] Uneducated (n=7) Some education (n=13) 12 Completely uneducated % larger responses 100 100 24 Received some education 80 80 w = 0.288 w = 0.177 60 60 “choose the larger” 40 40 * 20 20 0 0 0.7 1 1.4 0.7 1 1.4 Ratio of n1 and n2 (log scale) Weber fraction Munduruku, uneducated Weber fraction 0.5 Munduruku, some education 0.5 Italian participants (group means) r²=26.8%, p=0.001 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0.0 0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7 8 Age Years of Education
  • 60. Conclusions -There is some good evidence for a fundational role of the parietal system for approximate numerosity in symbolic numerical representations. But there is a lot to be discovered: 1) A true causal role of the ANS in dyscalculia awaits confirmation (longitudinal studies) 2) What are the neural mechanisms that drive the refinement of the quantity code for symbolic stimuli? Are they necessarily mediated by language? 3) Which aspects of maths education enhance approximate number prepresentation acuity?
  • 61. THREE PARIETAL CIRCUITS FOR NUMBER PROCESSING Left hemisphere Axial slice Right hemisphere A. x = - 48 z = 44 z = 49 x = 39 50 % HORIZONTAL SEGMENT OF THE INTRAPARIETAL SULCUS (hips) 22 % B. x = - 49 z = 30 x = 54 LEFT ANGUALR GYRUS C. x = - 26 z = 61 x = 12 POSTERIOR SUPERIOR PARIETAL LOBE (more right)
  • 62. Three parietal circuits for number processing (Dehaene, Piazza et al.,2003) CS Left hemisphere Right hemisphere IPS HORIZONTAL SEGMENT OF THE INTRAPARIETAL SULCUS (HIPS) hVIP? -Number comparison -Ratio effect Seen from top -Numerical priming -Approximate calculation LEFT ANGUALR GYRUS (l AG) -Retrieval of arithmetical facts (multiplications, additions) -Simple exact calculation POSTERIOR SUPERIOR PARIETAL LOBE (more right) (PSPL)  vLIP? -Subtractions -Complex additions -Approximate calculation
  • 63. Evidence for a verbal code in arithmetical facts retrieval • Interference on TRs in calculation Task1 (arithmetic): Multiplicazions or subtractions Task 2 (short term memory): Phonological (whisper a non- word) o visuo-spatial (remember the position of an object) Single task Phonological dual task Visuo-spatial dual task
  • 64. Left angular gyrus in arithmetical facts retrieval 1. Training experiment: Trained to memorize complex two digits number arithmetical facts and measure the effects on brain activity UNTRAINED > TRAINED TRAINED > UNTRAINED Ischebeck et al., 2009
  • 65. Left angular gyrus in arithmetical facts retrieval 2. Arithmetical tasks performed in the scanner and activation correlated with subsequent subjects’ report on the strategy used (fact retreival or computation) [Grabner et al., 2009 ]
  • 66. Evidence for a spatial code in arithmetical computations • Interference on TRs in calculation Task1 (arithmetic): Multiplicazions or subtractions Task 2 (short term memory): Phonological (whisper a non- word) o visuo-spatial (remember the position of an object) Single task Phonological dual task Visuo-spatial dual task
  • 67. Evidence for a spatial code in arithmetical computations Do spatial/motor processes interfere with calculation ? « Answer the arithmetical problems while performing a sequence of finger movements in the same time ! »
  • 68. NO MVTS MVTS * 1200 * 1100 1000 RT (msec) 900 800 700 600 MULTIPLICATION ADDITION SOUSTRACTION In the dual task, sequential finger movements were found to slow down responses to additions and subtractions, whereas multiplications (matched for difficulty) were unaffected
  • 69. Evidence for a spatial code in arithmetical computations: neglect Modello Copia del paziente Regioni corticali tipicamente dannegggiate nel neglect Tipico disegno (copia da modello) di un paziente con negelct Test della bisezione di linee Test dello sbarramento di linee
  • 70. Evidence for a spatial code in arithmetical computations: neglect Numerical bisection test : “What is the number between 2 and 6?” “Answer: 5”  RIGHT BIAS! Zorzi et al., Nature 2002
  • 72.
  • 73. Spatial code in number representations: the mental number line (SNARC effect) is culture dependent Shaki et al., 2009 (Psych Bull Rev)