docs: simplify SDIL deck around real figures

7 files changed, 262 insertions, 525 deletions

diff --git a/slides/SDIL_project_intro.pptx b/slides/SDIL_project_intro.pptx
index 5efde76..7544f97 100644
--- a/slides/SDIL_project_intro.pptx
+++ b/slides/SDIL_project_intro.pptx
diff --git a/slides/SDIL_project_intro.py b/slides/SDIL_project_intro.py
index 19e49c5..357f3db 100644
--- a/slides/SDIL_project_intro.py
+++ b/slides/SDIL_project_intro.py
@@ -1,55 +1,34 @@
 #!/usr/bin/env python3
-"""Build the context-free SDIL project introduction deck.
-
-All quantitative claims come from tracked project result files. The deck uses
-only completed full-method experiments: the controlled traffic panel, the
-standard ResNet-20/32/56 panel, and the calibrated synthetic BCI confirmation.
-"""
+"""Build a plain, evidence-first English introduction to the SDIL project."""
 
 from pathlib import Path
-from typing import Iterable, Sequence
 
 from PIL import Image
 from pptx import Presentation
 from pptx.dml.color import RGBColor
-from pptx.enum.shapes import MSO_CONNECTOR, MSO_SHAPE
 from pptx.enum.text import MSO_ANCHOR, PP_ALIGN
 from pptx.util import Inches, Pt
 
 
 ROOT = Path(__file__).resolve().parents[1]
-OUT_DIR = ROOT / "slides"
-ASSET_DIR = OUT_DIR / "assets"
-OUT_PATH = OUT_DIR / "SDIL_project_intro.pptx"
+SLIDES = ROOT / "slides"
+ASSETS = SLIDES / "assets"
+OUTPUT = SLIDES / "SDIL_project_intro.pptx"
 
 FIG3 = ROOT / "results/figs/figure3_innovation.png"
+FIG4 = ROOT / "results/figs/figure4_resnet_confirmation.png"
 FIG5 = ROOT / "results/figs/figure5_bci_v2.png"
+FIG6 = ROOT / "results/figs/figure6_standard_depth_scaling.png"
 
-SLIDE_W = 13.333
-SLIDE_H = 7.5
-
-FONT = "Noto Sans CJK SC"
-FONT_LATIN = "DejaVu Sans"
+FONT = "DejaVu Sans"
+BLACK = "111111"
+GRAY = "111111"
+LIGHT_GRAY = "B5B5B5"
+WHITE = "FFFFFF"
 
-BG = "F7F9FB"
-PAPER = "FFFFFF"
-INK = "17212B"
-MUTED = "5D6975"
-LIGHT = "DDE4EA"
-GRID = "D9E0E6"
-BLUE = "0072B2"
-BLUE_LIGHT = "DCEFF9"
-GREEN = "009E73"
-GREEN_LIGHT = "DDF3EC"
-ORANGE = "E69F00"
-ORANGE_LIGHT = "FBEBC9"
-RED = "D55E00"
-GRAY = "777777"
-DARK_GRAY = "3F4850"
 
-
-def rgb(hex_color: str) -> RGBColor:
-    return RGBColor.from_string(hex_color)
+def rgb(value: str) -> RGBColor:
+    return RGBColor.from_string(value)
 
 
 def add_text(
@@ -61,27 +40,25 @@ def add_text(
     h: float,
     *,
     size: float = 18,
-    color: str = INK,
     bold: bool = False,
+    color: str = BLACK,
     align=PP_ALIGN.LEFT,
     valign=MSO_ANCHOR.TOP,
-    margin: float = 0.03,
+    margin: float = 0.02,
     font: str = FONT,
-    line_spacing: float = 1.0,
 ):
     box = slide.shapes.add_textbox(Inches(x), Inches(y), Inches(w), Inches(h))
-    tf = box.text_frame
-    tf.clear()
-    tf.word_wrap = True
-    tf.margin_left = Inches(margin)
-    tf.margin_right = Inches(margin)
-    tf.margin_top = Inches(margin)
-    tf.margin_bottom = Inches(margin)
-    tf.vertical_anchor = valign
-    p = tf.paragraphs[0]
-    p.alignment = align
-    p.line_spacing = line_spacing
-    run = p.add_run()
+    frame = box.text_frame
+    frame.clear()
+    frame.word_wrap = True
+    frame.margin_left = Inches(margin)
+    frame.margin_right = Inches(margin)
+    frame.margin_top = Inches(margin)
+    frame.margin_bottom = Inches(margin)
+    frame.vertical_anchor = valign
+    paragraph = frame.paragraphs[0]
+    paragraph.alignment = align
+    run = paragraph.add_run()
     run.text = text
     run.font.name = font
     run.font.size = Pt(size)
@@ -90,72 +67,26 @@ def add_text(
     return box
 
 
-def add_rich_text(
-    slide,
-    runs: Sequence[tuple[str, float, str, bool]],
-    x: float,
-    y: float,
-    w: float,
-    h: float,
-    *,
-    align=PP_ALIGN.LEFT,
-    valign=MSO_ANCHOR.TOP,
-    margin: float = 0.03,
-):
+def add_bullets(slide, items, x, y, w, h, *, size=18, gap=7, color=BLACK):
     box = slide.shapes.add_textbox(Inches(x), Inches(y), Inches(w), Inches(h))
-    tf = box.text_frame
-    tf.clear()
-    tf.word_wrap = True
-    tf.margin_left = Inches(margin)
-    tf.margin_right = Inches(margin)
-    tf.margin_top = Inches(margin)
-    tf.margin_bottom = Inches(margin)
-    tf.vertical_anchor = valign
-    p = tf.paragraphs[0]
-    p.alignment = align
-    for text, size, color, bold in runs:
-        run = p.add_run()
-        run.text = text
-        run.font.name = FONT
-        run.font.size = Pt(size)
-        run.font.bold = bold
-        run.font.color.rgb = rgb(color)
-    return box
-
-
-def add_bullets(
-    slide,
-    items: Iterable[str],
-    x: float,
-    y: float,
-    w: float,
-    h: float,
-    *,
-    size: float = 17,
-    color: str = INK,
-    bullet_color: str = BLUE,
-    gap: float = 7,
-):
-    box = slide.shapes.add_textbox(Inches(x), Inches(y), Inches(w), Inches(h))
-    tf = box.text_frame
-    tf.clear()
-    tf.word_wrap = True
-    tf.margin_left = Inches(0.03)
-    tf.margin_right = Inches(0.03)
-    tf.margin_top = Inches(0.02)
-    tf.margin_bottom = Inches(0.02)
-    for idx, item in enumerate(items):
-        p = tf.paragraphs[0] if idx == 0 else tf.add_paragraph()
-        p.level = 0
-        p.space_after = Pt(gap)
-        p.line_spacing = 1.05
-        p.text = ""
-        marker = p.add_run()
-        marker.text = "●  "
-        marker.font.name = FONT
-        marker.font.size = Pt(max(size - 4, 10))
-        marker.font.color.rgb = rgb(bullet_color)
-        body = p.add_run()
+    frame = box.text_frame
+    frame.clear()
+    frame.word_wrap = True
+    frame.margin_left = Inches(0.02)
+    frame.margin_right = Inches(0.02)
+    frame.margin_top = Inches(0.01)
+    frame.margin_bottom = Inches(0.01)
+    for index, item in enumerate(items):
+        paragraph = frame.paragraphs[0] if index == 0 else frame.add_paragraph()
+        paragraph.space_after = Pt(gap)
+        paragraph.line_spacing = 1.05
+        paragraph.text = ""
+        bullet = paragraph.add_run()
+        bullet.text = "•  "
+        bullet.font.name = FONT
+        bullet.font.size = Pt(size)
+        bullet.font.color.rgb = rgb(color)
+        body = paragraph.add_run()
         body.text = item
         body.font.name = FONT
         body.font.size = Pt(size)
@@ -163,470 +94,273 @@ def add_bullets(
     return box
 
 
-def add_rect(
-    slide,
-    x: float,
-    y: float,
-    w: float,
-    h: float,
-    *,
-    fill: str = PAPER,
-    line: str = LIGHT,
-    radius: bool = True,
-    line_width: float = 1.0,
-):
-    kind = MSO_SHAPE.ROUNDED_RECTANGLE if radius else MSO_SHAPE.RECTANGLE
-    shape = slide.shapes.add_shape(kind, Inches(x), Inches(y), Inches(w), Inches(h))
-    shape.fill.solid()
-    shape.fill.fore_color.rgb = rgb(fill)
-    shape.line.color.rgb = rgb(line)
-    shape.line.width = Pt(line_width)
-    return shape
-
-
-def add_line(
-    slide,
-    x1: float,
-    y1: float,
-    x2: float,
-    y2: float,
-    *,
-    color: str = MUTED,
-    width: float = 1.5,
-    dash=None,
-):
-    line = slide.shapes.add_connector(
-        MSO_CONNECTOR.STRAIGHT,
-        Inches(x1),
-        Inches(y1),
-        Inches(x2),
-        Inches(y2),
-    )
-    line.line.color.rgb = rgb(color)
-    line.line.width = Pt(width)
-    if dash is not None:
-        line.line.dash_style = dash
-    return line
-
-
-def add_arrow(slide, x1, y1, x2, y2, *, color=MUTED, width=2.0):
-    line = add_line(slide, x1, y1, x2, y2, color=color, width=width)
-    line.line.end_arrowhead = True
-    return line
-
-
-def add_circle(slide, cx, cy, r, *, fill=BLUE, line=PAPER, line_width=1.0):
-    shape = slide.shapes.add_shape(
-        MSO_SHAPE.OVAL,
-        Inches(cx - r),
-        Inches(cy - r),
-        Inches(2 * r),
-        Inches(2 * r),
-    )
-    shape.fill.solid()
-    shape.fill.fore_color.rgb = rgb(fill)
-    shape.line.color.rgb = rgb(line)
-    shape.line.width = Pt(line_width)
-    return shape
-
-
 def add_title(slide, title: str, number: int, subtitle: str | None = None):
-    add_text(slide, title, 0.55, 0.28, 11.9, 0.54, size=26, bold=True)
-    add_text(slide, f"{number:02d}", 12.35, 0.34, 0.45, 0.3, size=11, color=MUTED, align=PP_ALIGN.RIGHT)
-    add_line(slide, 0.55, 0.91, 12.78, 0.91, color=LIGHT, width=1.0)
+    add_text(slide, title, 0.52, 0.28, 11.95, 0.55, size=27, bold=True)
+    add_text(slide, str(number), 12.35, 0.35, 0.42, 0.25, size=10, color=GRAY, align=PP_ALIGN.RIGHT)
     if subtitle:
-        add_text(slide, subtitle, 0.58, 1.00, 12.0, 0.35, size=13.5, color=MUTED)
+        add_text(slide, subtitle, 0.55, 0.92, 12.10, 0.36, size=13.5, color=GRAY)
 
 
 def add_footer(slide, text: str):
-    add_text(slide, text, 0.58, 7.18, 12.15, 0.2, size=8.5, color=MUTED, valign=MSO_ANCHOR.MIDDLE)
+    add_text(slide, text, 0.55, 7.17, 12.20, 0.18, size=8.5, color=GRAY)
 
 
-def add_badge(slide, text, x, y, w, *, fill=BLUE_LIGHT, color=BLUE, size=13):
-    add_rect(slide, x, y, w, 0.43, fill=fill, line=fill, radius=True)
-    add_text(slide, text, x + 0.05, y + 0.02, w - 0.1, 0.35, size=size, color=color, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
+def add_takeaway(slide, text: str, y: float = 6.55):
+    add_text(slide, text, 0.70, y, 11.93, 0.40, size=16, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
 
 
 def prepare_crops():
-    ASSET_DIR.mkdir(parents=True, exist_ok=True)
-    img = Image.open(FIG5)
-    w, h = img.size
-    crops = {
-        "figure5a_actor_critic.png": (0, 0, w // 2, h // 2),
-        "figure5d_outcome.png": (w // 2, h // 2, w, h),
-    }
-    for name, box in crops.items():
-        img.crop(box).save(ASSET_DIR / name)
-
-
-def add_custom_line_chart(
-    slide,
-    x: float,
-    y: float,
-    w: float,
-    h: float,
-    categories: Sequence[str],
-    series: Sequence[tuple[str, Sequence[float], str]],
-    *,
-    ymin: float,
-    ymax: float,
-    yticks: Sequence[float],
-    title: str,
-    show_legend: bool = True,
-):
-    add_rect(slide, x, y, w, h, fill=PAPER, line=LIGHT, radius=True)
-    add_text(slide, title, x + 0.25, y + 0.12, w - 0.5, 0.35, size=15, bold=True)
-    left = x + 0.60
-    right = x + w - 0.25
-    top = y + 0.72
-    bottom = y + h - 0.65
-
-    def px(i):
-        if len(categories) == 1:
-            return (left + right) / 2
-        return left + i * (right - left) / (len(categories) - 1)
-
-    def py(v):
-        return bottom - (v - ymin) * (bottom - top) / (ymax - ymin)
-
-    for tick in yticks:
-        yy = py(tick)
-        add_line(slide, left, yy, right, yy, color=GRID, width=0.8)
-        add_text(slide, f"{tick:g}", x + 0.08, yy - 0.12, 0.42, 0.24, size=9.5, color=MUTED, align=PP_ALIGN.RIGHT, valign=MSO_ANCHOR.MIDDLE)
-
-    add_line(slide, left, top, left, bottom, color=DARK_GRAY, width=1.1)
-    add_line(slide, left, bottom, right, bottom, color=DARK_GRAY, width=1.1)
-    for i, cat in enumerate(categories):
-        xx = px(i)
-        add_line(slide, xx, bottom, xx, bottom + 0.07, color=DARK_GRAY, width=1.0)
-        add_text(slide, cat, xx - 0.38, bottom + 0.10, 0.76, 0.28, size=9.5, align=PP_ALIGN.CENTER)
-
-    for name, values, color in series:
-        points = [(px(i), py(v)) for i, v in enumerate(values)]
-        for (x1, y1), (x2, y2) in zip(points, points[1:]):
-            add_line(slide, x1, y1, x2, y2, color=color, width=2.3)
-        for xx, yy in points:
-            add_circle(slide, xx, yy, 0.065, fill=color, line=PAPER, line_width=0.8)
-
-    if show_legend:
-        legend_y = y + h - 0.27
-        total = len(series)
-        slot = (w - 0.55) / total
-        for i, (name, _, color) in enumerate(series):
-            lx = x + 0.30 + i * slot
-            add_line(slide, lx, legend_y + 0.08, lx + 0.26, legend_y + 0.08, color=color, width=2.3)
-            add_circle(slide, lx + 0.13, legend_y + 0.08, 0.04, fill=color, line=color)
-            add_text(slide, name, lx + 0.32, legend_y - 0.04, slot - 0.34, 0.25, size=9.2, color=INK, valign=MSO_ANCHOR.MIDDLE)
+    ASSETS.mkdir(parents=True, exist_ok=True)
+
+    image5 = Image.open(FIG5)
+    w5, h5 = image5.size
+    image5.crop((0, 0, w5 // 2, h5 // 2)).save(ASSETS / "figure5a_actor_critic.png")
+    image5.crop((w5 // 2, h5 // 2, w5, h5)).save(ASSETS / "figure5d_outcome.png")
+
+    image6 = Image.open(FIG6)
+    w6, h6 = image6.size
+    image6.crop((0, 0, w6, h6 // 2)).save(ASSETS / "figure6_top_row.png")
 
 
 def build_deck():
     prepare_crops()
-    prs = Presentation()
-    prs.slide_width = Inches(SLIDE_W)
-    prs.slide_height = Inches(SLIDE_H)
-    blank = prs.slide_layouts[6]
+
+    presentation = Presentation()
+    presentation.slide_width = Inches(13.333)
+    presentation.slide_height = Inches(7.5)
+    blank = presentation.slide_layouts[6]
 
     def new_slide():
-        slide = prs.slides.add_slide(blank)
-        background = slide.background
-        background.fill.solid()
-        background.fill.fore_color.rgb = rgb(BG)
+        slide = presentation.slides.add_slide(blank)
+        slide.background.fill.solid()
+        slide.background.fill.fore_color.rgb = rgb(WHITE)
         return slide
 
-    # Slide 1: title and one-minute summary.
+    # 1. Title.
     slide = new_slide()
-    add_text(slide, "SDIL", 0.62, 0.60, 3.0, 0.72, size=38, color=BLUE, bold=True)
-    add_text(slide, "让局部学习使用意外的反馈", 0.62, 1.28, 7.1, 0.70, size=29, bold=True)
+    add_text(slide, "Somato-Dendritic Innovation Learning", 0.70, 1.02, 11.95, 0.75, size=34, bold=True)
+    add_text(slide, "Local credit assignment from soma-unpredicted dendritic signals", 0.72, 1.96, 11.75, 0.52, size=22)
     add_text(
         slide,
-        "Somato-Dendritic Innovation Learning\n受 Francioni et al.（Harnett lab）Nature 2026 启发",
-        0.65,
-        2.10,
-        6.5,
-        0.88,
+        "Inspired by the neuron-specific somato-dendritic residuals reported by Francioni et al., Nature 2026",
+        0.72,
+        2.70,
+        11.50,
+        0.55,
         size=16,
-        color=MUTED,
-        line_spacing=1.08,
+        color=GRAY,
+    )
+    add_text(
+        slide,
+        "Question: should a local learner use apical activity itself, or the component that is unexpected from the same neuron's somatic state?",
+        0.72,
+        4.35,
+        11.60,
+        1.00,
+        size=23,
+        bold=True,
     )
+    add_text(slide, "Project overview · August 2026", 0.72, 6.80, 4.0, 0.30, size=11, color=GRAY)
 
-    add_rect(slide, 7.55, 0.72, 5.05, 2.42, fill=PAPER, line=LIGHT, radius=True, line_width=1.2)
-    add_text(slide, "核心操作", 7.90, 1.02, 1.4, 0.35, size=14, color=MUTED, bold=True)
-    add_text(slide, "teaching signal", 7.92, 1.48, 1.75, 0.34, size=16, color=DARK_GRAY, align=PP_ALIGN.RIGHT)
-    add_text(slide, "=", 9.76, 1.45, 0.35, 0.40, size=22, color=MUTED, bold=True, align=PP_ALIGN.CENTER)
-    add_text(slide, "apical feedback", 10.12, 1.31, 2.05, 0.33, size=15.5, color=INK, bold=True, align=PP_ALIGN.CENTER)
-    add_text(slide, "− expected from soma", 10.12, 1.75, 2.05, 0.33, size=15.5, color=BLUE, bold=True, align=PP_ALIGN.CENTER)
-    add_text(slide, "每个神经元各自减去正常 soma–dendrite 耦合", 7.90, 2.43, 4.25, 0.38, size=13.5, color=MUTED, align=PP_ALIGN.CENTER)
-
-    cards = [
-        ("BP-free", "学习器内无反向计算图", BLUE_LIGHT, BLUE),
-        ("92.76%", "ResNet-56 + 4× predictable traffic", GREEN_LIGHT, GREEN),
-        ("1.33×", "BP MAC estimate", ORANGE_LIGHT, RED),
-    ]
-    for i, (big, small, fill, color) in enumerate(cards):
-        x = 0.65 + i * 4.12
-        add_rect(slide, x, 4.17, 3.75, 1.47, fill=fill, line=fill, radius=True)
-        add_text(slide, big, x + 0.18, 4.36, 3.39, 0.48, size=25, color=color, bold=True, align=PP_ALIGN.CENTER)
-        add_text(slide, small, x + 0.18, 4.91, 3.39, 0.38, size=12.3, color=INK, align=PP_ALIGN.CENTER)
+    # 2. Biological observation and learning rule.
+    slide = new_slide()
+    add_title(slide, "From the biological observation to a local learning rule", 2)
     add_text(
         slide,
-        "一句话：可扩展的局部信用路径负责把方向送到各层；SDIL 负责从混合反馈中提取可用于学习的部分。",
-        0.82,
-        6.30,
-        11.7,
-        0.50,
+        "The Harnett-lab study fits each neuron's normal soma–dendrite relationship. The residual is decorrelated from soma and carries outcome and signed task information.",
+        0.72,
+        1.18,
+        11.75,
+        0.76,
         size=18,
-        color=INK,
+    )
+    add_text(slide, "aₗ = sₗ + nₗ", 1.08, 2.25, 3.10, 0.52, size=25, bold=True, align=PP_ALIGN.CENTER)
+    add_text(slide, "mixed apical activity", 1.08, 2.78, 3.10, 0.32, size=13, color=GRAY, align=PP_ALIGN.CENTER)
+    add_text(slide, "âₗ = Pₗ(hₗ)", 5.10, 2.25, 3.10, 0.52, size=25, bold=True, align=PP_ALIGN.CENTER)
+    add_text(slide, "neutral-period prediction", 5.10, 2.78, 3.10, 0.32, size=13, color=GRAY, align=PP_ALIGN.CENTER)
+    add_text(slide, "rₗ = aₗ − âₗ", 9.13, 2.25, 3.10, 0.52, size=25, bold=True, align=PP_ALIGN.CENTER)
+    add_text(slide, "somato-dendritic innovation", 9.13, 2.78, 3.10, 0.32, size=13, color=GRAY, align=PP_ALIGN.CENTER)
+    add_text(
+        slide,
+        "ΔWₗ = η (rₗ ⊙ φ′(uₗ)) hₗ₋₁ᵀ",
+        2.63,
+        3.66,
+        8.10,
+        0.68,
+        size=29,
         bold=True,
         align=PP_ALIGN.CENTER,
     )
-    add_footer(slide, "Project introduction · ICLR 2027 work in progress · headline values use completed full experiments")
-
-    # Slide 2: scientific origin and problem.
-    slide = new_slide()
-    add_title(slide, "问题：反馈通道不只传教学信号", 2, "把全部 apical activity 当作 error，会把正常状态与上下文一起写入权重。")
-
-    add_rect(slide, 0.62, 1.52, 5.70, 4.88, fill=PAPER, line=LIGHT)
-    add_text(slide, "局部学习常见假设", 0.95, 1.78, 2.6, 0.38, size=17, bold=True)
-    add_rect(slide, 0.98, 2.42, 1.55, 0.76, fill=BLUE_LIGHT, line=BLUE)
-    add_text(slide, "teaching\nsignal", 1.10, 2.52, 1.31, 0.50, size=14, color=BLUE, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
-    add_rect(slide, 0.98, 3.62, 1.55, 0.76, fill="EFF1F3", line=GRAY)
-    add_text(slide, "ordinary\ntraffic", 1.10, 3.72, 1.31, 0.50, size=14, color=DARK_GRAY, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
-    add_arrow(slide, 2.55, 2.80, 3.55, 3.25, color=BLUE, width=2.1)
-    add_arrow(slide, 2.55, 4.00, 3.55, 3.55, color=GRAY, width=2.1)
-    add_rect(slide, 3.58, 2.72, 2.10, 1.36, fill="F3F5F6", line=DARK_GRAY)
-    add_text(slide, "raw apical\nfeedback", 3.76, 2.93, 1.74, 0.72, size=18, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
-    add_arrow(slide, 4.63, 4.10, 4.63, 4.85, color=RED, width=2.4)
-    add_rect(slide, 3.48, 4.90, 2.30, 0.84, fill="F8E6DE", line=RED)
-    add_text(slide, "update direction\nmay rotate", 3.65, 5.04, 1.96, 0.52, size=14.2, color=RED, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
-
-    add_rect(slide, 6.58, 1.52, 6.12, 4.88, fill=PAPER, line=LIGHT)
-    add_text(slide, "Nature 2026 提供了更具体的对象", 6.92, 1.78, 4.8, 0.38, size=17, bold=True)
-    # Simple soma-dendrite regression sketch.
-    plot_x, plot_y, plot_w, plot_h = 7.08, 2.60, 2.35, 2.30
-    add_line(slide, plot_x, plot_y + plot_h, plot_x + plot_w, plot_y + plot_h, color=DARK_GRAY, width=1.2)
-    add_line(slide, plot_x, plot_y + plot_h, plot_x, plot_y, color=DARK_GRAY, width=1.2)
-    add_line(slide, plot_x + 0.20, plot_y + 1.97, plot_x + 2.12, plot_y + 0.35, color=GRAY, width=2.0)
-    points = [(0.35, 1.74), (0.72, 1.41), (1.03, 1.29), (1.42, 0.88), (1.82, 0.61)]
-    for dx, dy in points:
-        add_circle(slide, plot_x + dx, plot_y + dy, 0.055, fill=GRAY, line=PAPER)
-    expected_x = plot_x + 1.45
-    expected_y = plot_y + 0.90
-    observed_y = plot_y + 0.35
-    add_circle(slide, expected_x, observed_y, 0.075, fill=BLUE, line=PAPER)
-    add_line(slide, expected_x, expected_y, expected_x, observed_y, color=BLUE, width=2.8)
-    add_text(slide, "residual", expected_x + 0.12, observed_y + 0.10, 0.80, 0.25, size=11.5, color=BLUE, bold=True)
-    add_text(slide, "soma", plot_x + 0.88, plot_y + plot_h + 0.18, 0.80, 0.25, size=10.5, color=MUTED, align=PP_ALIGN.CENTER)
-    add_text(slide, "dendrite", plot_x - 0.68, plot_y + 0.93, 0.70, 0.25, size=10.5, color=MUTED, align=PP_ALIGN.CENTER)
-
+    add_text(slide, "presynaptic activity × local postsynaptic gain × dendritic innovation", 2.62, 4.38, 8.12, 0.36, size=15, color=GRAY, align=PP_ALIGN.CENTER)
     add_bullets(
         slide,
         [
-            "先拟合每个神经元正常的 soma–dendrite 关系",
-            "残差与 soma 明显去相关",
-            "残差携带 outcome 与神经元特异的有符号任务信息",
+            "Training uses locally available quantities, forward observations, and manual synaptic updates.",
+            "Node-perturbation feedback supports the small-network experiments; reciprocal Kolen–Pollack plasticity supports the standard ResNets.",
+            "The inherited credit path transports layer-specific directions. SDIL removes soma-predictable traffic from those directions.",
         ],
-        9.75,
-        2.56,
-        2.62,
-        2.60,
-        size=14.3,
-        gap=8,
+        1.02,
+        5.05,
+        11.15,
+        1.45,
+        size=15.5,
+        gap=4,
     )
-    add_rect(slide, 6.93, 5.34, 5.42, 0.68, fill=BLUE_LIGHT, line=BLUE_LIGHT)
-    add_text(slide, "算法问题：学习是否也应使用 residual，而不是 raw activity？", 7.10, 5.50, 5.08, 0.33, size=15.2, color=BLUE, bold=True, align=PP_ALIGN.CENTER)
-    add_footer(slide, "Source: Francioni et al., “Vectorized instructive signals in cortical dendrites,” Nature (2026).")
+    add_footer(slide, "Sources: Francioni et al. (Nature 2026); Lansdell et al. (ICLR 2020); Akrout et al. (NeurIPS 2019).")
 
-    # Slide 3: method.
+    # 3. Controlled ablation.
     slide = new_slide()
-    add_title(slide, "方法：预测正常耦合，再用残差完成局部更新", 3, "SDIL 改变教学变量；它不要求重新发明整条反馈传播路径。")
-
-    steps = [
-        ("1", "Neutral observation", "âₗ = Pₗ(hₗ)", "估计同一神经元的正常耦合", "EFF1F3", DARK_GRAY),
-        ("2", "Innovation", "rₗ = aₗ − âₗ", "只保留 soma 无法预测的部分", BLUE_LIGHT, BLUE),
-        ("3", "Local plasticity", "ΔWₗ = η(rₗ ⊙ φ′(uₗ))hₗ₋₁ᵀ", "pre × post gain × dendritic residual", GREEN_LIGHT, GREEN),
-    ]
-    for i, (num, label, eq, desc, fill, color) in enumerate(steps):
-        x = 0.67 + i * 4.18
-        add_rect(slide, x, 1.70, 3.76, 2.38, fill=fill, line=color, radius=True, line_width=1.4)
-        add_circle(slide, x + 0.35, 2.04, 0.20, fill=color, line=color)
-        add_text(slide, num, x + 0.23, 1.90, 0.24, 0.28, size=13, color=PAPER, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE)
-        add_text(slide, label, x + 0.68, 1.83, 2.72, 0.36, size=16, color=color, bold=True)
-        add_text(slide, eq, x + 0.22, 2.48, 3.32, 0.53, size=19, color=INK, bold=True, align=PP_ALIGN.CENTER, valign=MSO_ANCHOR.MIDDLE, font=FONT_LATIN)
-        add_text(slide, desc, x + 0.30, 3.34, 3.16, 0.43, size=13.5, color=MUTED, align=PP_ALIGN.CENTER)
-        if i < 2:
-            add_arrow(slide, x + 3.82, 2.86, x + 4.10, 2.86, color=MUTED, width=2.0)
-
-    add_rect(slide, 0.67, 4.42, 5.96, 2.02, fill=PAPER, line=LIGHT)
-    add_text(slide, "学习器内部为什么是 BP-free", 0.98, 4.68, 3.2, 0.35, size=17, bold=True)
-    add_bullets(
+    add_title(
         slide,
-        [
-            "更新只读取 pre-synaptic activity、local gain 和本细胞 residual",
-            "无 reverse-mode autograd；不复制 forward-weight transpose",
-        ],
-        0.98,
-        5.12,
-        5.25,
-        1.12,
-        size=13.5,
-        gap=5,
+        "Residualization is load-bearing under predictable traffic",
+        3,
+        "Controlled MNIST intervention · depth 3 · width 256 · five paired seeds",
     )
+    slide.shapes.add_picture(str(FIG3), Inches(0.42), Inches(1.37), width=Inches(12.48), height=Inches(4.87))
+    add_takeaway(slide, "At traffic strength ρ = 0.5: raw 10.38%, norm-matched raw 10.31%, innovation 97.35%.")
+    add_footer(slide, "Norm matching equalizes per-example update magnitude. The remaining difference comes from teaching direction.")
 
-    add_rect(slide, 6.88, 4.42, 5.78, 2.02, fill=PAPER, line=LIGHT)
-    add_text(slide, "信用路径是可替换的已有组件", 7.19, 4.68, 3.8, 0.35, size=17, bold=True)
+    # 4. Standard ResNet-20 confirmation.
+    slide = new_slide()
+    add_title(
+        slide,
+        "Dynamic innovation survives a standard ResNet-20 test",
+        4,
+        "CIFAR-10 · 200 epochs · five untouched paired seeds · four-times-RMS soma-predictable traffic",
+    )
+    slide.shapes.add_picture(str(FIG4), Inches(0.43), Inches(1.47), width=Inches(12.46), height=Inches(3.82))
     add_bullets(
         slide,
         [
-            "Small nets: node-perturbation feedback vectorizer",
-            "ResNet: reciprocal KP plasticity",
-            "两者均为已有组件；SDIL 的贡献是 residualization",
+            "Dynamic SDIL under traffic: 91.584% mean test accuracy; clean reciprocal KP: 91.388%.",
+            "Mean early-layer teaching-signal cosine: 0.999687.",
+            "Cost: 1.326× the matched BP MAC estimate, zero task-loss queries, and one neutral observation per example.",
         ],
-        7.19,
-        5.10,
-        5.06,
-        1.18,
-        size=13.2,
-        bullet_color=GREEN,
+        1.20,
+        5.56,
+        10.95,
+        1.08,
+        size=15.5,
         gap=3,
     )
-    add_footer(slide, "Credit substrates: Lansdell et al. (ICLR 2020); reciprocal plasticity: Akrout et al. (NeurIPS 2019).")
+    add_footer(slide, "Exact gradients serve diagnostic measurement; local signals drive every learning update.")
 
-    # Slide 4: controlled causal evidence.
+    # 5. Standard-depth scaling.
     slide = new_slide()
-    add_title(slide, "关键消融：减法改变了方向，不只是信号大小", 4, "强 soma-predictable traffic 下，raw 与 norm-matched raw 都失效；innovation 保留学习方向。")
-    slide.shapes.add_picture(str(FIG3), Inches(0.40), Inches(1.38), width=Inches(12.52), height=Inches(4.89))
-    add_rect(slide, 1.02, 6.36, 11.25, 0.57, fill=BLUE_LIGHT, line=BLUE_LIGHT)
-    add_rich_text(
+    add_title(
         slide,
-        [
-            ("ρ = 0.5：", 15, INK, True),
-            ("raw 10.38%", 15, DARK_GRAY, True),
-            ("  ·  norm-matched raw 10.31%", 15, ORANGE, True),
-            ("  ·  innovation 97.35%", 15, BLUE, True),
-            ("  （5 seeds）", 13, MUTED, False),
-        ],
-        1.18,
-        6.47,
-        10.92,
-        0.33,
+        "Scaling from ResNet-20 to ResNet-56",
+        5,
+        "Complete 60-endpoint CIFAR-10 panel · five seeds per method and depth · shared hyperparameters",
+    )
+    figure6_top = ASSETS / "figure6_top_row.png"
+    slide.shapes.add_picture(str(figure6_top), Inches(0.56), Inches(1.42), width=Inches(12.22), height=Inches(4.38))
+    add_text(
+        slide,
+        "SDIL under 4× traffic: 91.584% → 92.254% → 92.760%; all five R20/R56 pairs improve.",
+        0.73,
+        6.01,
+        11.90,
+        0.40,
+        size=14.8,
+        bold=True,
+        align=PP_ALIGN.CENTER,
+    )
+    add_text(
+        slide,
+        "Reciprocal KP provides the scalable credit path. SDIL preserves this path under mixed apical traffic.",
+        0.73,
+        6.48,
+        11.90,
+        0.34,
+        size=14.5,
         align=PP_ALIGN.CENTER,
-        valign=MSO_ANCHOR.MIDDLE,
     )
-    add_footer(slide, "Controlled MNIST traffic intervention. Norm matching uses the same per-example update magnitude as innovation.")
+    add_footer(slide, "BP and clean reciprocal KP follow the same depth trend; fixed DFA remains near 31%.")
 
-    # Slide 5: standard-scale evidence.
+    # 6. Dynamical task.
     slide = new_slide()
-    add_title(slide, "标准 ResNet：扩展性来自 KP，抗混合流量来自 SDIL", 5, "60 个 CIFAR-10 endpoints；每个深度 5 seeds，200 epochs，无 depth-specific tuning。")
-
-    categories = ["R20", "R32", "R56"]
-    bp = [91.624, 92.302, 92.632]
-    dfa = [31.878, 32.684, 30.850]
-    kp = [91.388, 92.332, 92.670]
-    sdil = [91.584, 92.254, 92.760]
-    add_custom_line_chart(
+    add_title(
         slide,
-        0.62,
-        1.48,
-        5.92,
-        4.82,
-        categories,
-        [("BP", bp, DARK_GRAY), ("DFA", dfa, ORANGE), ("clean KP", kp, GREEN), ("SDIL + traffic", sdil, BLUE)],
-        ymin=25,
-        ymax=95,
-        yticks=[30, 50, 70, 90],
-        title="完整尺度：固定 DFA 随深度仍处于约 31%",
+        "A local actor–critic produces temporal outcome signals",
+        6,
+        "Synthetic BCI task · six task clusters × five model seeds · manual local updates",
     )
-    add_custom_line_chart(
+    panel_a = ASSETS / "figure5a_actor_critic.png"
+    panel_d = ASSETS / "figure5d_outcome.png"
+    slide.shapes.add_picture(str(panel_a), Inches(0.58), Inches(1.42), width=Inches(5.96), height=Inches(4.43))
+    slide.shapes.add_picture(str(panel_d), Inches(6.80), Inches(1.42), width=Inches(5.96), height=Inches(4.43))
+    add_text(
+        slide,
+        "Final success: 100% · terminal outcome decoding: 99.83% · outcome-lesion effect: 0.400.",
+        0.68,
+        6.06,
+        11.98,
+        0.36,
+        size=14.8,
+        bold=True,
+        align=PP_ALIGN.CENTER,
+    )
+    add_text(
         slide,
-        6.80,
-        1.48,
-        5.92,
-        4.82,
-        categories,
-        [("BP", bp, DARK_GRAY), ("clean KP", kp, GREEN), ("SDIL + 4× traffic", sdil, BLUE)],
-        ymin=91.0,
-        ymax=93.0,
-        yticks=[91.0, 91.5, 92.0, 92.5, 93.0],
-        title="近 BP 放大：SDIL 在 4× traffic 下跟随 clean KP",
+        "The synthetic task supplies terminal reward and tests neuron-specific outcome surprise with causal lesions.",
+        0.68,
+        6.52,
+        11.98,
+        0.32,
+        size=14.0,
+        align=PP_ALIGN.CENTER,
     )
-    add_badge(slide, "SDIL: 91.58 → 92.76%", 0.88, 6.45, 3.42, fill=BLUE_LIGHT, color=BLUE)
-    add_badge(slide, "1.31–1.33× BP MAC", 4.48, 6.45, 3.42, fill=ORANGE_LIGHT, color=RED)
-    add_badge(slide, "0 loss queries · 1 neutral obs/example", 8.08, 6.45, 4.25, fill=GREEN_LIGHT, color=GREEN, size=12.2)
-    add_footer(slide, "Source: complete 60-record ResNet panel. SDIL-specific claim is robustness under traffic, not clean superiority over KP.")
+    add_footer(slide, "Outcome labels and exact causal roles serve evaluation.")
 
-    # Slide 6: dynamical task evidence.
-    slide = new_slide()
-    add_title(slide, "动态任务：residual 携带 outcome surprise，而不只是分类误差", 6, "独立的 synthetic BCI actor–critic：6 task clusters × 5 models，所有学习更新均为手写局部规则。")
-    p_a = ASSET_DIR / "figure5a_actor_critic.png"
-    p_d = ASSET_DIR / "figure5d_outcome.png"
-    slide.shapes.add_picture(str(p_a), Inches(0.57), Inches(1.46), width=Inches(5.95), height=Inches(4.42))
-    slide.shapes.add_picture(str(p_d), Inches(6.80), Inches(1.46), width=Inches(5.95), height=Inches(4.42))
-    add_badge(slide, "100% final task success", 0.93, 6.07, 3.44, fill=BLUE_LIGHT, color=BLUE)
-    add_badge(slide, "99.83% terminal outcome decoding", 4.73, 6.07, 3.82, fill=GREEN_LIGHT, color=GREEN, size=12.4)
-    add_badge(slide, "outcome lesion: −0.400 separation", 8.92, 6.07, 3.50, fill=ORANGE_LIGHT, color=RED, size=12.0)
-    add_text(slide, "范围：这是机制验证用的合成任务；terminal reward 被直接提供，不是皮层数据。", 0.82, 6.66, 11.70, 0.32, size=13.5, color=MUTED, align=PP_ALIGN.CENTER)
-    add_footer(slide, "Source: complete untouched calibrated BCI confirmation; outcome labels and exact roles are diagnostic-only.")
-
-    # Slide 7: positioning and next decisive evidence.
+    # 7. Current scientific position.
     slide = new_slide()
-    add_title(slide, "当前最准确的定位：可扩展 local credit 的混合信号分离模块", 7)
-
-    # Pipeline strip.
-    pipeline = [
-        ("existing credit path", "KP / learned feedback", "EFF1F3", DARK_GRAY, 0.66, 2.28),
-        ("mixed apical channel", "instruction + ordinary traffic", ORANGE_LIGHT, RED, 3.33, 2.76),
-        ("SDIL", "subtract soma-predictable part", BLUE_LIGHT, BLUE, 6.52, 2.52),
-        ("local update", "eligibility × innovation", GREEN_LIGHT, GREEN, 9.48, 2.80),
-    ]
-    for i, (head, sub, fill, color, x, w) in enumerate(pipeline):
-        add_rect(slide, x, 1.35, w, 0.98, fill=fill, line=color, radius=True, line_width=1.2)
-        add_text(slide, head, x + 0.12, 1.51, w - 0.24, 0.28, size=14.2, color=color, bold=True, align=PP_ALIGN.CENTER)
-        add_text(slide, sub, x + 0.12, 1.87, w - 0.24, 0.24, size=10.8, color=INK, align=PP_ALIGN.CENTER)
-        if i < len(pipeline) - 1:
-            next_x = pipeline[i + 1][4]
-            add_arrow(slide, x + w + 0.06, 1.84, next_x - 0.07, 1.84, color=MUTED, width=1.8)
-
-    columns = [
-        (0.66, 3.05, 3.80, "已经建立", BLUE, BLUE_LIGHT, [
-            "可预测 traffic 会旋转 raw local update",
-            "per-neuron residualization 在受控干扰下是 load-bearing 的",
-            "组合方法在 ResNet-20/32/56 上保持近 BP accuracy",
-        ]),
-        (4.76, 3.05, 3.80, "真正的新东西", GREEN, GREEN_LIGHT, [
-            "把 Harnett residual 直接定义成 teaching variable",
-            "neutral-period predictor + dynamic local projection",
-            "机制、方向、成本和动态 outcome 的联合证据",
-        ]),
-        (8.86, 3.05, 3.80, "仍然缺少", RED, ORANGE_LIGHT, [
-            "自然任务中不可预先测量的 mixed feedback",
-            "真实硬件或真实神经数据上的必要性",
-            "证明 SDIL 的价值不能由 clean KP 单独解释",
-        ]),
-    ]
-    for x, y, w, head, color, fill, items in columns:
-        add_rect(slide, x, y, w, 2.68, fill=PAPER, line=LIGHT)
-        add_rect(slide, x, y, w, 0.56, fill=fill, line=fill)
-        add_text(slide, head, x + 0.18, y + 0.10, w - 0.36, 0.32, size=16, color=color, bold=True, align=PP_ALIGN.CENTER)
-        add_bullets(slide, items, x + 0.25, y + 0.78, w - 0.50, 1.70, size=13.2, bullet_color=color, gap=7)
-
-    add_rect(slide, 1.06, 6.10, 11.20, 0.67, fill=BLUE_LIGHT, line=BLUE_LIGHT)
+    add_title(slide, "Current scientific position", 7)
+    add_text(slide, "Established", 0.78, 1.30, 3.30, 0.38, size=20, bold=True)
+    add_bullets(
+        slide,
+        [
+            "Soma-predictable apical traffic can rotate a raw local teaching signal.",
+            "Per-neuron innovation preserves learning direction under controlled traffic.",
+            "Dynamic innovation works with a scalable BP-free reciprocal credit path on ResNet-20/32/56.",
+            "A separate local actor–critic reproduces neuron-specific outcome signaling in a controlled dynamical task.",
+        ],
+        0.82,
+        1.82,
+        11.50,
+        2.15,
+        size=17,
+        gap=7,
+    )
+    add_text(slide, "Attribution", 0.78, 4.15, 3.30, 0.38, size=20, bold=True)
+    add_bullets(
+        slide,
+        [
+            "Reciprocal KP provides clean-setting scale and feedback tracking.",
+            "SDIL provides soma-conditioned traffic removal and the associated mechanism tests.",
+        ],
+        0.82,
+        4.66,
+        11.50,
+        0.92,
+        size=17,
+        gap=7,
+    )
+    add_text(slide, "Next decisive experiment", 0.78, 5.82, 4.20, 0.38, size=20, bold=True)
     add_text(
         slide,
-        "下一项决定性证据：在自然或硬件产生的 state-dependent mixed traffic 中，raw credit 失败，而 SDIL 无需 oracle calibration 即可恢复。",
-        1.27,
-        6.25,
-        10.78,
-        0.34,
-        size=14.6,
-        color=BLUE,
-        bold=True,
-        align=PP_ALIGN.CENTER,
+        "Evaluate raw reciprocal credit and SDIL under naturally generated, state-dependent mixed traffic with component-level bias hidden from the learner.",
+        0.82,
+        6.30,
+        11.45,
+        0.58,
+        size=18,
     )
-    add_footer(slide, "Bounded conclusion: strong controlled mechanism + standard-scale compatibility; natural mixed-traffic necessity remains open.")
+    add_footer(slide, "Target claim: scalable local credit assignment with adaptive removal of predictable feedback contamination.")
 
-    OUT_DIR.mkdir(parents=True, exist_ok=True)
-    prs.save(OUT_PATH)
-    print(OUT_PATH)
+    SLIDES.mkdir(parents=True, exist_ok=True)
+    presentation.save(OUTPUT)
+    print(OUTPUT)
 
 
 if __name__ == "__main__":
diff --git a/slides/qa-ledger.md b/slides/qa-ledger.md
index 3375b35..33fa5c2 100644
--- a/slides/qa-ledger.md
+++ b/slides/qa-ledger.md
@@ -6,3 +6,6 @@
 | Method title and three-stage visual used inconsistent step counts | PPTX/PDF | 3 | medium | Removed the numeric step count from the title | resolved |
 | Chinese font, formula glyph, chart and image rendering | PPTX/PDF | all | high | Exported through LibreOffice and inspected all seven rendered slides | resolved |
 | Claim attribution | PPTX/PDF | 5 and 7 | high | States that clean scaling comes from reciprocal KP and SDIL-specific evidence is traffic robustness | resolved |
+| First deck used decorative cards and redrawn charts | PPTX/PDF | all | high | Rebuilt the deck in English with white background, black text, and tracked experimental figures | resolved |
+| Rebuttal-style negative sentences | PPTX/PDF | all | medium | Rephrased attribution and scope as direct affirmative statements | resolved |
+| Long title and summary lines wrapped into adjacent content | PPTX/PDF | 5 and 6 | high | Shortened the title and summary statements while preserving every reported value | resolved |
diff --git a/slides/rendered/.gitignore b/slides/rendered/.gitignore
index 82ae779..d73e6f7 100644
--- a/slides/rendered/.gitignore
+++ b/slides/rendered/.gitignore
@@ -1,2 +1,3 @@
 slide-[0-9]*.png
 slide3_check.png
+qa-*.png
diff --git a/slides/rendered/SDIL_project_intro.pdf b/slides/rendered/SDIL_project_intro.pdf
index 0378cdb..38406c0 100644
--- a/slides/rendered/SDIL_project_intro.pdf
+++ b/slides/rendered/SDIL_project_intro.pdf
diff --git a/slides/rendered/contact_sheet.png b/slides/rendered/contact_sheet.png
index ad0ddc2..0b17b1f 100644
--- a/slides/rendered/contact_sheet.png
+++ b/slides/rendered/contact_sheet.png
diff --git a/slides/visual-contract.md b/slides/visual-contract.md
index 86a7ffc..26fd3d6 100644
--- a/slides/visual-contract.md
+++ b/slides/visual-contract.md
@@ -1,14 +1,13 @@
 # SDIL project introduction deck: visual contract
 
-- **Artifact:** Seven-slide, 16:9 project introduction deck.
-- **Audience:** Machine-learning researchers with no prior SDIL context.
-- **Core claim:** A local learner should use the soma-unpredicted component of a mixed apical signal; this residualization protects an inherited scalable credit path from predictable traffic.
-- **Reader questions:** Why is raw feedback insufficient? What exactly is new? Is the update BP-free? Is residualization necessary? Does the combined method scale? Which evidence is controlled or synthetic?
-- **Evidence layers:** problem and mechanism (slides 2–3), causal ablation (slide 4), standard-depth scaling and cost (slide 5), dynamical-task evidence (slide 6), attribution and boundary (slide 7).
-- **Source data:** `results/figs/figure3_innovation.png`, `results/figs/figure5_bci_v2.png`, `results/oral_a_dynamic_scaling_v2_gate.json`, and the evidence-bound manuscript.
-- **Statistics:** Five seeds for the controlled traffic and ResNet panels; six task clusters by five model seeds for the synthetic BCI panel. Values shown are audited means or paired outcomes already reported in the manuscript.
-- **Visual grammar:** Direct mechanism diagrams, two result figures, and editable line charts. SDIL is blue, reciprocal KP is green, DFA is orange, BP is dark gray.
+- **Artifact:** Seven-slide, 16:9 English project introduction deck.
+- **Audience:** Machine-learning researchers new to SDIL.
+- **Core claim:** A local learner can use the soma-unpredicted component of a mixed apical signal; residualization protects an inherited scalable credit path from predictable traffic.
+- **Reader questions:** What problem does mixed feedback create? How does the learning rule work? Is residualization load-bearing? Does the combined system scale? Which component supplies each capability?
+- **Evidence layers:** biological motivation and rule (slide 2), causal ablation (slide 3), standard-ResNet confirmation (slide 4), standard-depth scaling (slide 5), dynamical-task evidence (slide 6), attribution and next experiment (slide 7).
+- **Source figures:** `results/figs/figure3_innovation.png`, `results/figs/figure4_resnet_confirmation.png`, `results/figs/figure5_bci_v2.png`, and `results/figs/figure6_standard_depth_scaling.png`.
+- **Statistics:** Five seeds for the controlled traffic and ResNet panels; six task clusters by five model seeds for the synthetic BCI panel.
+- **Visual grammar:** White background, black text, plain typography, full experimental figures, and cropped experimental panels. One equation slide carries the method. The deck adds zero process diagrams and zero redrawn result charts.
 - **Exact method labels:** SDIL, raw apical signal, somatic prediction, innovation, reciprocal KP, BP, DFA, local update, neutral observation.
 - **Output:** Editable PPTX, PDF export, generation source, and rendered QA contact sheet.
-- **Claim boundary:** Clean scaling is attributed to reciprocal KP; SDIL-specific evidence is robustness under predictable mixed traffic and the synthetic BCI mechanism test.
-
+- **Attribution:** Reciprocal KP supplies clean-setting scale. SDIL supplies soma-conditioned traffic removal and the associated mechanism evidence.