- Explain how limited observation and methodological choices affect scientific understanding.
- Distinguish proximate explanations from ultimate explanations in biology.
- Describe the evidence used to investigate pair living and paternal care in owl monkeys.
- Identify major limits of observational evidence, proxy measurements, small samples, missing data, and binary p-value thresholds.
- Separate statistical description from causal inference and population-level prediction from individual prediction.
- A basic understanding of biological evolution.
- Familiarity with observation, hypotheses, and scientific evidence.
Universities as places for shared inquiry
The workshop joins philosophical questions about understanding with empirical methods from several disciplines.
The workshop began from a question shared across disciplines: how can researchers draw sound inferences while recognizing the weaknesses of their methods? Bringing different scholarly languages into conversation can reveal both common problems and field-specific limitations.
Tamar Gendler described the university as an institution that preserves, produces, and passes on knowledge. This workshop exemplified that mission by linking fundamental philosophical work on understanding with practical investigation in empirical fields.
The event also showed how interdisciplinary work can be sustained in practice. It developed over several years, used live and recorded formats, connected unexpected groups, combined outside and campus resources, and relied on existing networks that made the project visible.
Partial observation and the search for knowledge
Hume's swans illuminate the limits of observation, the contingency of experience, and the influence of method.
Hume's swan example exposes an asymmetry in reasoning: repeated sightings of white swans cannot establish that every swan is white, while one black swan can overturn that claim. The example invites reflection on what observation can and cannot justify.
- Ask whether an apparently superficial feature rests on a deep structure, or whether something that looks fundamental is only a local manifestation. This question appears in fields ranging from gender and sexuality studies to engineering.
- Remember that ordinary and unusual are defined by experience. Had Hume lived among Australia's many black swans, the universal claim about white swans might never have suggested itself.
- Consider how inquiry changes what becomes visible. Ethnographic approaches may delay theory to remain receptive, while theory-led approaches warn that inquiry without a framework may miss what matters.
Both receptive observation and theory-guided investigation can contribute to knowledge. The methodological variety of interdisciplinary research matters because different approaches disclose different aspects of a problem.
Two kinds of biological explanation
Proximate and ultimate causes answer different but complementary versions of why.
In a 1961 paper, evolutionary biologist Ernst Mayr distinguished proximate causes from ultimate causes. The distinction provides the organizing framework for examining causation in biological anthropology.
The older account pictured causal influence as a largely bottom-up sequence from genome to cell, organism, and environment, followed by inheritance into the next generation. Later work replaced this one-way picture with reciprocal influence across levels.
Under the extended evolutionary synthesis, genomes influence cells and organisms, while cellular, organismal, and environmental processes also matter. Changes during an individual's lifetime may affect a later generation. This view moves beyond the gene-centered picture associated with Richard Dawkins's The Selfish Gene.
How does a proximate explanation differ from an ultimate explanation?
A proximate account identifies the mechanisms producing a phenomenon, while an ultimate account addresses its adaptive or functional evolutionary significance.
Working with attainable evidence
Biological anthropology combines biological and cultural perspectives while accepting that ideal data are rarely available.
Biological anthropology, also called evolutionary anthropology or physical anthropology, studies what being human means when biology and evolution are taken into account. Human life cannot be understood without culture, but humans are also organisms composed of cells and genes.
Researchers investigate broad questions across long spans of space and time: what separates humans from chimpanzees, when cooperation over food emerged, when fire was used for warmth or cooking, and when walking on two legs began.
- Fossils and artifacts can be physically examined, although researchers may still dispute who produced a particular tool.
- Genetic information can help reconstruct the past and clarify relationships among individuals.
- Behavior does not fossilize. Researchers must observe living animals or reason from traces such as ancient footprints.
A footprint in an African savanna is dated to 1.3 million years ago. What could it support directly, and where would behavioral interpretation begin to exceed the trace itself?
The footprint directly supports the presence and movement of an organism at that place and time. Richer claims about its broader behavior require additional evidence and careful inference.
Owl monkeys as a case in causal reasoning
Long-term field research illustrates the difficulty of explaining pair living, social monogamy, and paternal care.
Eduardo's research asks two connected questions: why do some primates live in pairs and remain socially monogamous, and why do some males provide extensive paternal care? The goal is to report findings while making their uncertainty and limitations visible.
A long-running field study
Research teams have studied owl monkeys in Ecuador, Peru, and Argentina, alongside captive colonies in Texas and California. Moving between laboratory control and natural settings allows the same questions to be approached in different ways.
The main Argentine study area covers roughly 200 hectares, or 500 acres. Fieldworkers use telemetry collars and accelerometers, observe behavior directly, gather feces and sometimes tissue, record life histories and demography, and measure environmental variables.
Continuous year-round work began in May 1996. The accumulated record spans 28 years and now includes a fourth generation of identified animals, creating opportunities to connect behavior, life history, genetics, and environmental conditions.
Why remain in one pair?
A prominent evolutionary model links female distribution to food, predators, and pathogens, then connects male strategies to infanticide risk, opportunities for paternal care, competition among males, and the distribution of receptive females. Possible social outcomes include pairs, solitary females visited by roaming males, one male with several females, or multi-male and multi-female groups.
The Argentine forest contains about 18 owl-monkey groups. A typical group has one adult male and one adult female within a territory near five hectares, about 12 acres. Some pairs range across twice that area, showing that a male can cover approximately ten hectares.
Proximate inquiry: care, energy, and cortisol
Paternal care is especially puzzling in mammals because fertilization occurs inside the female reproductive tract, creating uncertainty about fatherhood, while pregnancy and lactation impose major energetic demands on the female.
Within days of birth, the male owl monkey becomes the main caregiver apart from nursing. Males carry infants, share food, groom, touch, play, and respond when an infant calls. Laboratory observations also suggest that infants attach more strongly to fathers than to mothers.
One hypothesis proposes that male care lowers the female's energetic burden by allowing her to rest or sleep. Researchers therefore examine cortisol as a metabolic gatekeeper connected with glucose metabolism rather than treating it only as a stress hormone.
The desired evidence would concern cortisol and its receptors in the brain, but researchers cannot obtain that measurement from these wild animals. Blood and urine are also unavailable, so the feasible method is to collect feces and ask whether their contents provide an adequate proxy.
Ultimate inquiry: survival and paternity
Owl-monkey infants survive at unusually high rates for wild primates. One possible explanation is that care from both parents improves survival, although the observed association does not by itself settle the causal question.
Evolutionary interpretation also requires evidence about paternity. Researchers distinguish social monogamy, sexual monogamy, and genetic monogamy rather than assuming that living together proves exclusive reproduction.
Among 120 tested pairs of offspring and putative fathers, no offspring was assigned to another male. Paternity assessment works by excluding incompatible candidates rather than directly confirming a father's identity, but this record still indicates an extreme system of genetic monogamy.
A male committed to one relationship may therefore sire young who may themselves tend toward care, pair living, and pair bonding. Pair bonding could be necessary for fatherhood and paternal care to evolve, but the evidence does not show that it is sufficient.
If fecal cortisol were your only available energetic measure, what additional evidence or comparison would strengthen an inference about paternal care?
Use multiple kinds of evidence, compare field findings with controlled research where possible, examine whether the proxy tracks relevant biological states, and state clearly what remains unmeasured.
Unresolved problems in statistical and causal reasoning
Observational research requires explicit attention to sample size, effect meaning, missingness, prediction targets, and the difference between association and cause.
Biological anthropology relies mainly on observational data and usually has few subjects. Researchers may also lack a principled account of which effect sizes matter biologically, as in the unresolved question of what difference in fecal cortisol would matter to an owl monkey.
Beyond the p-value cutoff
Eduardo argued that null-hypothesis significance testing was inadequate for much of his research. He pointed to the American Statistical Association's 2016 statement on p-values and a later Nature paper signed by one hundred authors as signs of growing concern.
Noise, bias, and absent observations
Noise and bias should not be treated as interchangeable labels for undesirable data. They are distinct problems and need separate analysis.
Missing observations also demand explanation. Researchers should ask why particular data were not obtained and consider how those causes affect inference instead of silently passing over the omissions.
Population patterns are not individual forecasts
Researchers must state whether a model concerns a population or predicts an individual. Public audiences may hear that a behavior can be modeled and wrongly assume that the result will determine what a particular individual does.
Statistical inference is not causal inference
Students often blur the boundary between data and what is inferred from them. The broader workshop problem is equally important: a statistical pattern does not automatically reveal its cause.
When direct experiments are unavailable, researchers can strengthen inference through controlled laboratory comparisons, comparative biology, genetics, and abduction: reasoning toward the explanation that best fits several bodies of evidence.
Integrating the lessons
Strong inquiry combines disciplinary depth, methodological plurality, and candor about uncertainty.
- Observation is always limited by where an observer stands and by the information available there.
- Mechanistic and evolutionary explanations address different levels of the same biological problem.
- The extended evolutionary synthesis treats causal influence across genome, cell, organism, and environment as reciprocal.
- Scientific maturity includes naming the distance between obtainable evidence and ideal evidence.
- An arbitrary p-value boundary should not determine whether a finding is worth thinking about or reporting.
- Keep data separate from inference, social monogamy from genetic monogamy, and population models from individual forecasts.
- Cross-disciplinary training can reveal tools that a field has not yet incorporated into its routine practice.
Eduardo's own path moved from neurobiology of learning and memory to experimental psychology, biological anthropology, statistics, and causal-inference methods. That movement across fields shaped both his questions and his willingness to reconsider established research habits.
What assumption in your own field might arise from a narrow sample of the world, and what observation could challenge it?
For owl-monkey paternal care, identify one proximate question, one ultimate question, and one limitation on the available evidence.
A proximate question asks whether male care reduces the female's energetic burden and how cortisol relates to that process. An ultimate question asks whether care improves infant survival and is favored when males sire the young. A major limitation is reliance on observational data and indirect fecal measurements.
Common misconceptions
Watch out: Cortisol should be interpreted only as a stress hormone.
Instead: In this research it is treated more broadly as a metabolic gatekeeper related to glucose use, while fecal cortisol remains an uncertain proxy for the desired brain measurement.
Watch out: If two animals live together, they must reproduce exclusively with one another.
Instead: Social, sexual, and genetic monogamy are separate claims. Genetic evidence is needed to assess paternity.
Watch out: Observed closeness proves that wild owl monkeys possess a bond in the experimentally demonstrated sense.
Instead: The wild observations support pair-living language, but the Argentine study has not directly shown distress after separation or active reunion seeking. Stronger bond evidence comes from related captive research systems.
Watch out: Crossing the p < 0.05 threshold establishes that a result is important.
Instead: An arbitrary cutoff does not determine biological meaning or causal value. Interpretation must return to domain knowledge, effect meaning, study limitations, and careful reasoning.
Lesson recap
Scientific understanding grows through partial observations interpreted with explicit methods. Biological explanations need both proximate mechanisms and ultimate functions, while long-term owl-monkey research shows how difficult it is to connect behavior, energetics, paternity, survival, and evolution.
Reliable inquiry does not hide imperfect proxies, small samples, missing observations, or unresolved causal direction. It distinguishes what was measured from what was inferred and relies on disciplinary judgment rather than a single statistical threshold.