Addition and Subtraction by Human Infants Study Notes

Theoretical Foundations of Infant Numerical Cognition

  • Core Research Question: Investigating whether human infants possess true numerical concepts capable of calculating simple arithmetic operations, or if infant numerical discrimination is driven entirely by non-numerical, holistic perceptual pattern recognition.

  • Baseline Infant Numerical Capabilities:

    • Human infants demonstrate the ability to discriminate between small numbers of items (ranging from 11 to 44 items).
    • Infants can determine numerical equivalence across different perceptual modalities (such as matching visual set sizes to auditory output).
  • Theoretical Debate on the Nature of Subitization:

    • Subitization Definition: The cognitive mechanism by which small numbers of items are quantified rapidly without conscious, explicit counting.
    • Pattern-Recognition Model: Asserts that subitization relies on the holistic recognition of canonical perceptual patterns. Under this framework, subitization yields non-numerical percepts that fail to encode ordinal or arithmetical relationships between numbers.
    • Iterative/Counting Model: Asserts that subitization is an iterative or counting-like mechanism that explicitly specifies numerical and ordinal relationships between items.
  • Core Empirical Findings:

    • 55-month-old human infants can calculate the precise results of simple arithmetical additions (1+11 + 1) and subtractions (2−12 - 1).
    • Demonstrates that infants possess true numerical concepts and suggests humans are innately endowed with basic arithmetical calculation abilities.
    • Establishes that subitization is a process that encodes ordinal information rather than a pattern-recognition process yielding non-numerical percepts.

Experimental Paradigm and Methodological Controls

  • Violation-of-Expectation Looking-Time Paradigm:

    • Standard cognitive methodology based on the principle that infants look significantly longer at unexpected or physically impossible events than at expected or possible events.
    • If infants can compute the numerical results of arithmetical operations, they will exhibit longer visual gaze durations when the outcome of an operation is mathematically incorrect.
  • Physical Apparatus and Manipulations:

    • Display Setup: A display case featuring an empty display area and a rotating screen that lifts up to hide items from view and drops down to reveal final outcomes.
    • Trap Door Mechanism: Manipulations of objects behind the screen are performed by an experimenter hidden behind the rear wall of the display, utilizing a hidden trap door.
    • Occlusion Rule: Infants observe the full process of the arithmetical operation (adding or removing objects), but cannot see the final result until the screen drops.
  • Experimental Control Measures:

    • Observer Blindness: A hidden observer timing the infant's visual gaze is kept strictly unaware of the infant's group assignment (1+11 + 1 vs. 2−12 - 1) and trial ordering.
    • Parental Behavior: Parents either stand out of sight behind the infant without touching them or gently touch the infant while facing completely away from the display case to prevent behavioral cueing.
  • Participant Selection and Exclusion Criteria:

    • Fussiness/Drowsiness: A total of 1616 infants were excluded across the experimental series due to becoming excessively fussy or drowsy.
    • Statistical Outliers: A total of 11 infant was excluded because their test preference exceeded 2.52.5 standard deviations from the mean for that group.
    • Pretest Baseline Cutoff: A total of 1919 infants were excluded for exhibiting a pretest preference of more than 10 s10\,s for either display number (re-analyses demonstrate that the inclusion of these infants does not alter the pattern or statistical significance of the results).

Experiment 1: Basic Addition (1+11 + 1) and Subtraction (2−12 - 1)

  • Subject Demographics:

    • Sample size: N=32N = 32 normal, full-term human infants.
    • Age range: 44 months 1919 days to 55 months 1616 days.
    • Mean age: 55 months 11 day.
    • Group distribution: Randomly divided into two equal groups (n=16n = 16 for the 1+11 + 1 group; n=16n = 16 for the 2−12 - 1 group).
  • Pretest Baseline Phase:

    • Infants were presented with a display containing 11 item and a display containing 22 items prior to arithmetic trials.
    • Looking times were recorded to establish individual baseline preferences for the two visual displays.
  • Test Phase Event Sequences:

    • Sequence for 1+1=11 + 1 = 1 or 22:
    1. A single object is placed in the empty display case in plain view.
    2. The screen rotates up, hiding the object.
    3. The experimenter brings a second identical object into the display area in clear view.
    4. The experimenter places the second item behind the screen out of the infant's sight; the hand leaves empty.
    5. The screen drops to reveal either the possible outcome of 22 objects or the impossible outcome of 11 object.
    • Sequence for 2−1=12 - 1 = 1 or 22:
    1. Two objects are placed in the display case in plain view.
    2. The screen rotates up, hiding both objects.
    3. The experimenter's empty hand enters behind the screen.
    4. The experimenter removes one object from behind the screen in clear view; the hand leaves holding the object.
    5. The screen drops to reveal either the possible outcome of 11 object or the impossible outcome of 22 objects.
  • Trial Administration and Termination Rules:

    • Each infant received 66 test trials, alternating between 11-item and 22-item outcomes.
    • Trial order was counterbalanced: half received (1,2,1,2,1,2)(1, 2, 1, 2, 1, 2) and half received (2,1,2,1,2,1)(2, 1, 2, 1, 2, 1).
    • End-of-Trial Criterion: A trial concluded when an infant looked away for 22 consecutive seconds after looking at the display for at least 44 cumulative seconds, or reached 3030 cumulative seconds of looking.
  • Quantitative Results (Experiment 1):

    • Preference metric calculation: P(2)=LT(2)−LT(1)P(2) = LT(2) - LT(1), where LT(1)LT(1) and LT(2)LT(2) represent mean looking times in seconds for 11 item and 22 items.
    • Pretest Baseline Data:
    • 1+11 + 1 Group: LT(1)=20.06 sLT(1) = 20.06\,s, LT(2)=20.80 sLT(2) = 20.80\,s, P(2)=0.74 sP(2) = 0.74\,s.
    • 2−12 - 1 Group: LT(1)=17.99 sLT(1) = 17.99\,s, LT(2)=19.61 sLT(2) = 19.61\,s, P(2)=1.62 sP(2) = 1.62\,s.
    • Statistical test: d.f.=30d.f. = 30, t=0.649t = 0.649, p>0.5p > 0.5 (22-tailed between-group t-test; no baseline preference difference).
    • Test Trial Data:
    • 1+11 + 1 Group: LT(1)=13.36 sLT(1) = 13.36\,s, LT(2)=12.80 sLT(2) = 12.80\,s, P(2)=−0.53 sP(2) = -0.53\,s (longer gaze at impossible result of 11 item).
    • 2−12 - 1 Group: LT(1)=10.54 sLT(1) = 10.54\,s, LT(2)=13.73 sLT(2) = 13.73\,s, P(2)=3.19 sP(2) = 3.19\,s (longer gaze at impossible result of 22 items).
    • Statistical test: d.f.=30d.f. = 30, t=2.078t = 2.078, p<0.05p < 0.05 (11-tailed between-group t-test).
    • Outcome: Both groups looked significantly longer at the incorrect outcomes than at the correct outcomes.

Experiment 2: Replication and Procedural Modification

  • Subject Demographics:

    • Sample size: N=16N = 16 infants (n=6n = 6 in the 1+11 + 1 group; n=10n = 10 in the 2−12 - 1 group).
    • Age range: 44 months 1818 days to 55 months 55 days.
    • Mean age: 44 months 2525 days.
  • Procedural Modification:

    • Criteria for end-of-trial modified such that the minimum required cumulative looking time was reduced from 4 s4\,s to 2 s2\,s (looking away for 22 consecutive seconds after at least 22 cumulative seconds, or 30 s30\,s maximum).
    • This procedural change explains the shorter overall mean looking times in Experiment 2 relative to Experiment 1.
  • Quantitative Results (Experiment 2):

    • Pretest Baseline Data:
    • 1+11 + 1 Group: LT(1)=11.12 sLT(1) = 11.12\,s, LT(2)=10.62 sLT(2) = 10.62\,s, P(2)=−0.50 sP(2) = -0.50\,s.
    • 2−12 - 1 Group: LT(1)=10.35 sLT(1) = 10.35\,s, LT(2)=11.44 sLT(2) = 11.44\,s, P(2)=1.09 sP(2) = 1.09\,s.
    • Statistical test: d.f.=14d.f. = 14, t=0.677t = 0.677, p>0.5p > 0.5 (22-tailed between-group t-test).
    • Test Trial Data:
    • 1+11 + 1 Group: LT(1)=12.08 sLT(1) = 12.08\,s, LT(2)=9.45 sLT(2) = 9.45\,s, P(2)=−2.65 sP(2) = -2.65\,s.
    • 2−12 - 1 Group: LT(1)=10.98 sLT(1) = 10.98\,s, LT(2)=8.05 sLT(2) = 8.05\,s, P(2)=2.94 sP(2) = 2.94\,s.
    • Statistical test: d.f.=14d.f. = 14, t=1.795t = 1.795, p<0.05p < 0.05 (11-tailed between-group t-test).
  • Combined Analysis of Experiments 1 and 2 (N=48N = 48):

    • Combined Pretest Data:
    • 1+11 + 1 Group: LT(1)=17.62 sLT(1) = 17.62\,s, LT(2)=18.02 sLT(2) = 18.02\,s, P(2) = 0.41\,s$.\n - 2 - 1Group:Group:LT(1) = 15.05\,s,,LT(2) = 16.47\,s,,P(2) = 1.42\,s$.
    • Statistical test: d.f.=46d.f. = 46, t=0.873t = 0.873, p>0.35p > 0.35 (22-tailed between-group t-test).
    • Combined Test Trial Data:
    • 1+11 + 1 Group: LT(1)=13.01 sLT(1) = 13.01\,s, LT(2)=11.89 sLT(2) = 11.89\,s, P(2) = -1.11\,s$.\n - 2 - 1Group:Group:LT(1) = 9.59\,s,,LT(2) = 12.67\,s,,P(2) = 3.09\,s$.
    • Statistical test: d.f.=46d.f. = 46, t=2.73t = 2.73, p<0.005p < 0.005 (11-tailed between-group t-test).
  • General Habituation Dynamics:

    • Across all test trials, looking times were systematically lower than pretest times because infants' gaze decreases over time as they grow familiar with the display.

Experiment 3: Precise Numerical Calculation vs. Directionless Expectation

  • Theoretical Distinction and Competing Hypotheses:

    • Experiments 1 and 2 established that infants expect arithmetical operations to cause numerical changes, but left open two hypotheses:
    1. Precise Calculation Hypothesis: Infants compute the exact numerical results of simple additions and subtractions (1+1=21 + 1 = 2; 2−1=12 - 1 = 1).
    2. Directionless Change Hypothesis: Infants expect operations to cause numerical change, but have no expectations regarding the specific direction or magnitude of the change (expecting 1+1≠11 + 1 \neq 1 and 2−1≠22 - 1 \neq 2 without knowing the target result).
  • Experimental Logic and Setup:

    • Tested 1+11 + 1 addition where the outcome revealed behind the screen was either 22 items (correct) or 33 items (incorrect).
    • Since both 22 and 33 items are numerically distinct from the initial single item (11), the directionless change hypothesis predicts equal looking times.
    • If infants compute precise numerical outcomes (1+1=21 + 1 = 2), they will be surprised by and look significantly longer at 33 items than 22 items.
  • Subject Demographics and Parameters:

    • Sample size: N=16N = 16 infants.
    • Age range: 44 months 44 days to 55 months 44 days.
    • Mean age: 44 months 1818 days.
    • Procedure: Used the same end-of-trial criterion as Experiment 2 (2 s2\,s minimum cumulative looking time).
  • Quantitative Results (Experiment 3):

    • Preference metric calculation: P(3)=LT(3)−LT(2)P(3) = LT(3) - LT(2), where LT(3)LT(3) and LT(2)LT(2) represent mean looking times in seconds to 33 items and 22 items.
    • Pretest Baseline Data:
    • LT(2)=14.16 sLT(2) = 14.16\,s, LT(3)=13.87 sLT(3) = 13.87\,s, P(3) = -0.29\,s$.\n - Statistical test against null hypothesis (P(3) = 0):):d.f. = 15,,t = -0.224,,p > 0.5((2-tailed t-test).\n - *Test Trial Data*:\n - LT(2) = 9.96\,s,,LT(3) = 11.89\,s,,P(3) = 1.92\,s$.
    • Statistical test against null hypothesis (P(3)=0P(3) = 0): d.f.=15d.f. = 15, t=2.044t = 2.044, p<0.03p < 0.03 (11-tailed t-test).
  • Conclusion: Infants showed a statistically significant preference for 33 items during test trials, demonstrating surprise when addition resulted in 33 items and supporting the claim that 55-month-olds compute exact numerical results.

Theoretical Evaluation and Alternative Hypotheses

  • Discrete Item Calculation vs. Continuous Quantity Measurement:

    • Continuous Substance Hypothesis: Proposes that infants calculate results based on continuous amounts of physical substance or total matter rather than discrete counts of individual items.
    • Rebuttal and Supporting Evidence for Discrete Entity Quantification:
    1. Empirical literature demonstrates infant sensitivity to small discrete numerical changes, but lacks evidence showing sensitivity to minor variations in physical matter volume.
    2. Infants are perceptually predisposed to interpret the physical world as composed of discrete, individual entities when processing spatial layouts.
    3. Infants explicitly represent the precise spatial locations and trajectories of individual physical objects relative to one another over space and time.
    4. The concept of an "individual entity" forms a foundational pillar in early conceptualization, supporting the existence of a specialized cognitive mechanism for quantifying collections of discrete entities.
  • Implications for Early Cognitive Architecture:

    • Human infants possess genuine numerical concepts with direct mental access to the ordering and numerical relationships of small numbers.
    • Subitization is confirmed to be a mechanism that generates true numerical outputs encoding quantitative relationships, rather than non-numerical holistic percepts derived from spatial pattern matching.
    • Early presence in 55-month-old infants suggests human beings are innately endowed with elementary arithmetical calculation capacities, providing the foundational cognitive substrate for developing complex mathematical knowledge.