Concepts and simple questions

Question 1

FPSmax normalization

Answer 1

Dividing ERF amplitudes by maximum signal during frequency-tagging

Question 2

N170

Answer 2

Adult ERP component occurring ~170ms after face presentation, sensitive to conscious face perception. Posterior topography.

Question 3

N290

Answer 3

Infant equivalent of N170, occurring at ~290ms (70% slower due to immature axonal conduction)

Question 4

FPS

Answer 4

Fast Periodic Stimulation with frequency-tagging approach

Particularly well-suited for immature nervous systems precisely because it doesn’t require precise temporal dynamics ;-)

Question 5

Permutation clustering

Answer 5

I proposed this to control multiple comparisons while identifying temporal clusters

Question 6

Active inference

Answer 6

Perception as predictive process combining bottom-up and top-down signals

Question 7

WP1

Answer 7

Pilot study estimating effect sizes (n=60)

Question 8

WP2

Answer 8

Main replication with larger sample (n=120-160)

Question 9

WP3

Answer 9

Heart rate responses to social stimuli

Question 10

WP4

Answer 10

oMEG validation against sMEG

Question 11

WP5

Answer 11

Postnatal follow-up with same subjects

Question 12

Cortical subplate

Answer 12

The cortical subplate is a transient but crucial layer of neurons that forms beneath the developing cerebral cortex during embryonic and early postnatal brain development. It plays several essential roles in cortical formation and connectivity.

Question 13

Regionally focal response hypothesis in infants

Answer 13

C5 (left, vocal) and P8 (right, faces), Z-test yielding z = 1.64, P < 0.05

Question 14

Earliest gestational age for recruitment

Answer 14

32 weeks gestation

Question 15

Time the total session takes for each participant

Answer 15

90 minutes

Question 16

Visual intensity of the face-like stimulus

Answer 16

12,500 lux [each LED is 5 lumens]. lux = light flux with units lumen per square meter

Question 17

Do any fetuses visit the lab twice?

Answer 17

Yes, for WP4, half of the participants from Wp2-3 will be tested with both SQUID-MEG and OPM-MEG at separate lab visits in randomized order.

Question 18

Strategies to calm the infant?

Answer 18

We will blow soap bubbles to relax the infant until the infant achieves a calm, relaxed state

Question 19

Experiment 1

Answer 19

WP1, 30 minutes total: 20 minutes of ERF trials (10 minutes per stimulus in random order) followed by 10 minutes of FPS (5 minutes per stimulus type in random order)

Question 20

Experiment 2

Answer 20

WP 2-4, 25 minutes total: 10 minutes of long-duration heart rate trials followed by 10 minutes of ERF trials (one stimulus per participant) and 5 minutes of frequency-tagging.

Question 21

Experiment 3

Answer 21

WP5, 12.5 - 25 minutes max: 7.5 minutes of ERFs with voice or face photo stimuli, 7.5 minutes of blurred face-like stimuli (if applicable), 5 minutes of frequency tagging with voice or face photos, 5 minutes of frequency tagging with blurred face-like stimuli (if applicable).

Question 22

MEG preprocessing

Answer 22

FLORA (heart detection), FAUAN (heart artifact removal), 1 - 15 Hz FIR filtering (fetal and neonatal), -200 to +1500 ms epoching

Question 23

MEG preprocessing specific to neonatal data

Answer 23

Artifact block algorithm by Mourad et al. [learn this], 4 STD channel rejection threshold, 2 pT rejection threshold.

Question 24

Statistical models

Answer 24

Linear (regression) models with gestational age as a covariate.

My team will use a log-likelihood ratio tests to assess the appropriateness of adding additional factors to
linear models. For example, if biological sex improves the model fit, we will incorporate the additional factor
as a covariate.

Otherwise, unnecessary model complexity will be avoided to prevent overfitting, using stepwise
variable selection if necessary. Furthermore, in the interest of reducing model complexity, my team will fit
separate models for each group of fetuses (forward-facing and control) in the FACE experiments, with the
expectation that only the forward-facing group will yield an effect. To account for multiple testing, we will
control the false discovery rate (FDR) in each WP using the Benjamini-Hochberg correction.

Question 25

Total personnel costs

Answer 25

Roughly 1.1 million EUR 1,098,660 EUR PI (80% time): €400,800 (2024 DFG junior group leader rate) PhD (75%, years 1–4): €255,360 Postdoc (100%, years 2–5): €344,400 Tech assistant (50%, years 1–3): €98,100 (for MEG data collection)

Question 26

Travel costs

Answer 26

31,400 EUR One to two major international conferences will be attended each year by the PI, PhD student, and postdoc (2,000 EUR per person per year) + 5-day research visit (Year 2) to Little Rock (1,800 EUR per person, PI, Postdoc, PhD students).

Question 27

Consumable costs

Answer 27

24,000 EUR 10K for light stim device + 14K for participant remuneration (20 EUR/hour)

Question 28

Publication costs

Answer 28

40,000 EUR

Question 29

Other costs

Answer 29

About 5.5K 5,696 EUR GitHub accounts (45 EUR per person per year) + 5000 EUR for the project audit.

Question 30

Heart rate variability

Answer 30

It's complicated and depends on HF and LF components. See email thread with Julia.

Question 31

FOXP2

Answer 31

FOXP2 is a gene found in many vertebrates, where it plays an important role in mimicry in birds (such as birdsong) and echolocation in bats. FOXP2 is also required for the proper development of speech and language in humans. In humans, mutations in FOXP2 cause the severe speech and language disorder developmental verbal dyspraxia. Studies of the gene in mice and songbirds indicate that it is necessary for vocal imitation and the related motor learning.

Question 32

Ephrins

Answer 32

Eph/ephrin signaling regulates a variety of biological processes through morphogen gradients during embryonic development including the guidance of axon growth cones, formation of tissue boundaries, cell migration, and segmentation.

Question 33

Netrins

Answer 33

A class of proteins involved in axon guidance. Netrins are chemotropic; a growing axon will either move towards or away from a higher concentration of netrin.

Question 34

What is your senior author paper about?

Question 35

Primiparous

Answer 35

pregnant for the first time "Pri -mip - erus"

Question 36

Fetal actocardiogram

Answer 36

Echocardiogram acquired during fetal stress. Some fetal stress testing uses non-exercise methods like maternal position changes, controlled contractions (contraction stress test), or acoustic stimulation of the fetus to evaluate how well the fetal heart responds to various stressors.

Question 37

echocardiogram

Answer 37

Uses ultrasound waves to create visual images of the heart's structure, chambers, valves, and blood flow patterns.

Question 38

How does the artifact blocking algorithm for infant MEG data work?

Answer 38

The Artifact-Blocking algorithm by Mourad et al. automatically removes high-amplitude EEG artifacts by first creating a "target" signal that zeros out samples above a threshold, then calculating a blocking matrix B that learns to transform the original contaminated signal to match this target as closely as possible. Crucially, the output during artifacts is not simply flat zeros, but rather a corrected EEG signal where B has learned to suppress the specific spatial patterns associated with artifacts (like frontal electrode combinations during eye blinks) while preserving the spatial patterns containing brain signals that occur simultaneously. This spatial selectivity allows the algorithm to subtract out artifact components while maintaining underlying neural activity, producing a cleaned signal that retains useful EEG data even during contaminated time periods.

Question 39

M cells vs P cells

Answer 39

M (magnocellular) LGN neurons have large receptive fields, respond transiently to stimuli, and project primarily to the dorsal "where/how" stream (toward parietal cortex) for spatial processing and action guidance. Especially relevant to face processing is the fact that they are respond to low spatial frequencies and stimuli in the peripheral visual field. P (parvocellular) LGN neurons have smaller receptive fields, sustained responses, and project primarily to the ventral "what" stream (toward temporal cortex) for object recognition.

Question 40

Is the magnocellular pathway relevant?

Answer 40

Yes, it responds to low spatial frequencies and stimuli in the periphery, meaning these LGN neurons can detect the face-like stimulus in the experiment. They carry information dorsal "where" stream, and also to the superior colliculus and pulvinar.

Question 41

Conspecific detection

Answer 41

The ability to identify and preferentially respond to members of one's own species versus other animals or objects. This is a fundamental biological capacity because conspecifics are typically the most important social partners for survival, mating, learning, and cooperation.

Question 42

How is heart rate tested in WP3?

Answer 42

20 trials each of social and control stimuli with 3-second stimulus duration 5-second baseline period before each stimulus, followed by 7-second post-stimulus interval Total trial length is 15 seconds (5s baseline + 3s stimulus + 7s post-stimulus) Instantaneous heart rate is computed as the reciprocal of each RR-interval between heartbeats Expected temporal resolution of roughly 500ms (since fetal heart rate is typically 120+ bpm) Baseline correction: Mean heart rate during the 5-second baseline is subtracted from the entire time series