Sensory ecology studies how organisms obtain information from their surroundings and use it to find food, avoid danger, navigate and interact. Its gold standard is a complete explanation: identify the environmental information, show how it reaches a particular receiver, test the receiver’s response and establish what that response means for the organism in its habitat.
This guide focuses on animals and on the connection between physical stimuli and ecological decisions. It includes original fictional calculations, not measured species thresholds or environmental design standards. A brighter object, a louder call or a stronger smell does not automatically provide better information. The useful question is what this animal can distinguish, in this setting, for this task.
Choose your sensory ecology route
Follow information from the environment to a biological decision, then test what that decision means.
Route 1 · Chapters 1–3: Follow the information
Route 2 · Chapters 4–6: Calculate what reaches the receiver
Route 3 · Chapters 7–8: Examine real research
Route 4 · Chapters 9–11: Test and repair explanations
Route 5 · Chapters 12–14: Practise and continue
Chapter index
Follow the information · Chapters 1–3
Calculate what reaches the receiver · Chapters 4–6
Examine real research · Chapters 7–8
Test and repair explanations · Chapters 9–11
Practise and continue · Chapters 12–14
A human looking across a pond sees water, plants, reflections and perhaps an insect. Another organism encounters a different set of usable contrasts. A surface may provide a directional cue, a vibration may indicate nearby movement, and a chemical trace may persist after its source has gone. The physical world is shared, but the information available from it depends on the receiver and the circumstances.
The University of Exeter’s sensory ecology research theme connects light, sound and chemicals with behaviour, communication and ecological interactions. That connection defines this subject’s distinctive question. We are not simply listing sense organs. We are asking how a sensory system makes a particular ecological task possible, and when that system becomes unreliable.
Start with a concrete task: locate a food source, maintain upright flight, distinguish a mate’s display or avoid an approaching predator. Then state the available cue and the action it might guide. “The animal sees well” is too broad. “The animal can discriminate this moving target from this background at this distance” is a claim that a study can test.
The same sensitivity can have different consequences across tasks. Detecting a faint stimulus may help an animal locate food, but it may also increase responses to irrelevant events. Remaining alert can reduce missed threats while taking time away from feeding. Sensory performance therefore belongs inside a decision problem, with benefits, errors and costs.
For the wider comparison of biological mechanisms, use Comparative Physiology. Here we follow the information that those mechanisms make available. Keep a five-part notebook entry: task, stimulus, receiver, response and ecological consequence. If one part is missing, record the uncertainty rather than completing it with an attractive story.
A source produces or reflects a physical stimulus. The environment modifies it. A receptor responds to some of the arriving energy or chemicals. Nervous processing combines that response with other information and the animal’s state. Behaviour changes what the animal encounters next. This last step matters: sensing is often active, because moving, turning or sampling changes the evidence available.
OpenStax’s account of sensory processes explains transduction: a receptor converts a suitable stimulus into a biological signal. Receptor activation and perception are different stages. A measurable response at a receptor does not, on its own, demonstrate recognition, a particular experience or a decision to act.
Consider a hypothetical experiment in which a fish approaches a coloured panel. At least four explanations remain: the panel resembles food; it contrasts strongly with the background; it changes the apparent shelter available; or an uncontrolled reflection accompanies the colour change. Naming the panel’s colour does not select among these mechanisms.
The repair is to vary features deliberately. Compare panels with different spectra while controlling intensity as appropriately as possible for the receiver. Change background and viewing angle. Measure approach, inspection and feeding attempts separately. A single endpoint called “interest” hides important differences between detection, investigation and attempted use.
Also distinguish an instrument’s success from the animal’s success. A microphone can record a sound outside a particular animal’s hearing range. A camera can create a bright image by integrating light for much longer than the animal’s visual system can use during rapid movement. Calibration helps relate the instrument to the environment; biological evidence is still needed to relate it to the receiver.
A cue supplies usable information without necessarily having evolved to communicate that information to the observer. A prey animal’s movement can reveal its position to a predator even though revealing itself is disadvantageous. A communication signal has an evolutionary role involving its effect on receivers. The distinction concerns function and history, not whether a human observer finds the pattern meaningful.
Keep the definition explicit when reading research because fields sometimes use “signal” more broadly for any measured input. An acoustic engineer’s signal can be a recorded waveform; a behavioural biologist’s communication signal involves a different claim. Neither usage authorises treating every detectable event as a message deliberately sent.
The existing Insect Communication guide develops pheromones, sound, vibration, light and receiver evidence. Sensory ecology also includes information that no other animal sends: illumination gradients, environmental motion and the structure of a background. These can guide behaviour without a sender–receiver conversation.
Information can also become misleading. A rule that worked in one environment may produce an inappropriate action after the environment changes. That does not require an animal to be unintelligent. A reliable relationship between a cue and the world may have broken. The scientific task is to identify the relationship and test the proposed failure mechanism.
Use precise language in a field note. Write “turned toward the flashing source in eight of twelve trials” before writing “recognised a mate”. The first statement records an observation. The second needs controls against other possible reasons for turning, such as general attraction to intermittent light. Clear language makes the next experiment easier to design.
An object’s appearance depends on illumination, its optical properties, the transmission path, the background and the receiver. A photograph of a bright red object does not establish that another species sees a bright red object in its natural habitat. Camera processing, flash illumination and human display settings can all change the comparison.
NOAA’s light and colour investigation explains why red sunlight disappears at shallower depths than blue light in the ocean. The lesson is not that every underwater setting has one fixed visibility depth. Particles, dissolved materials, depth, direction and local lighting also matter. A colour label alone leaves out the path by which light reaches an eye.
For an original simplified exercise, suppose a small target has measured luminance 12 units against a uniform background of 10 units. Define Weber contrast as C = (target − background)/background. The contrast is (12 − 10)/10 = 0.20, or 20%. The units cancel because both measurements use the same scale.
Now suppose both measurements double: the target becomes 24 and the background 20. Contrast remains 0.20. The target is physically brighter, but its relative contrast has not increased. Detection could nevertheless change if photon availability or adaptation matters. The calculation isolates one property; it does not predict every part of visual performance.
In a third condition, the target stays at 12 while the background rises to 11. Contrast becomes 1/11, approximately 0.091. That is less than half the original contrast. An intervention that raises background illumination can therefore change visibility even if the object itself is unchanged. This is an algebraic demonstration, not a measured animal threshold.
The definition has limits. It requires a nonzero reference background and is most interpretable for a target against a reasonably uniform field. Patterned scenes, moving targets and colour discrimination need richer descriptions. Do not insert a human luminance value into an animal colour model and assume that it measures the animal’s experience.
A call has a source, a propagation path and a receiver. Making a call louder can extend its physical reach under some conditions, but the receiver must still detect the relevant frequency pattern against background sound. Direction, timing, hearing sensitivity and competing sound all influence the result.
The Bioacoustics root covers recordings, propagation, masking and the limits of acoustic inference. Here the focus is the animal’s usable information space: the region in which an incoming sound can support a specific task. Detecting that something happened may require less information than identifying a caller or locating it accurately.
For a fictional worked example, assume a compact source in an ideal free field. Let its sound-pressure level be 60 dB at 1 metre. Ignore reflections, absorption, directionality and intervening obstacles. The model is L(r) = 60 − 20 log10(r/1 metre). The Simon Fraser University acoustic handbook describes the approximately 6 dB decrease for each doubling of distance under ideal conditions.
At 2 metres the model gives approximately 54 dB; at 4 metres, 48 dB; at 8 metres, 42 dB. Suppose, purely for this exercise, that reliable performance requires a signal 6 dB above a 30 dB background in the relevant frequency band. The required received level is 36 dB. Solving 60 − 20 log10(r) = 36 gives r = 10^1.2, approximately 15.85 metres.
If the background rises to 36 dB, the assumed required level becomes 42 dB. The predicted radius becomes 10^0.9, approximately 7.94 metres. A 6 dB background increase has approximately halved the modelled range. For an unobstructed circular area, the area ratio is (7.94/15.85)², approximately 0.251: about one quarter remains.
This is not evidence that any real species loses exactly three quarters of its communication area. The chosen margin is invented, real hearing is frequency dependent, and usable habitat is not necessarily a circle. The calculation demonstrates why a modest-looking decibel change can have a substantial geometric consequence under explicitly stated assumptions.
An odour source and an odour measurement are not the same location. Transport through air or water changes the concentration and timing experienced by a receiver. A single average concentration can conceal bursts separated by intervals with little detectable material. A useful investigation therefore records where and when the sample was taken, not just what chemical was present.
Before interpreting an animal’s path as movement along a smooth gradient, ask whether the environment actually supplied one. Airflow, obstacles and the release pattern may make the encounter sequence irregular. A laboratory arrangement with steady flow can answer a controlled question without representing every field condition. Keep that distinction visible when moving from a chamber to a landscape.
Mechanical information likewise depends on a path. A vibration measured at one point on a support need not match the vibration arriving at an animal elsewhere on it. The support’s material, connections and geometry belong in the explanation. Changing a support to make an experiment easier to film may also change the stimulus that the animal receives.
An accessible reasoning exercise uses no animals. Imagine a sensor beside an intermittently operated scent source. One minute contains a high reading followed by a long gap; another contains a lower but continuous reading. The same minute-average does not establish identical information for an animal responding to brief encounters. Ask which time scale its decision requires before choosing an averaging interval.
These examples encourage a common discipline across senses: specify the relevant spatial and temporal scale. A map of average conditions can be useful for broad planning while missing the brief local events that guide an immediate turn. Neither measurement is automatically wrong; they answer different questions.
In their 2024 study Why flying insects gather at artificial light, Fabian and colleagues used three-dimensional flight observations to examine behaviour around lights. Their results support a dorsal-light-response explanation: insects tilt their upper surface toward the light, and a nearby artificial source can produce persistent steering that traps them around it.
The study offers a more specific mechanism than simply saying insects “like light”. It connects body orientation with flight trajectories. Its scope should remain explicit: evidence about flight close to experimental lights does not settle every question about long-distance attraction, every insect group or every lighting configuration.
Use the case to construct competing predictions. If an explanation depends on the direction of an apparent upper light region, changing the light’s position should matter. If an alternative depends only on warmth, controlling heat while changing visual geometry becomes relevant. These are hypotheses for appropriately designed research, not instructions to experiment on wildlife at home.
The broader lesson is about intervention. “Less bright” describes only one dimension of a lamp. Direction, placement, duration and spectral output may change the information available. A mitigation proposal should specify which mechanism it addresses and which outcomes it will measure. Counting insects at a light and measuring their survival are not interchangeable evaluations.
For a classroom discussion, draw two candidate explanations and identify a measurement that would distinguish them. Do not ask students to memorise a dramatic replacement story as if one paper removed all uncertainty. Ask them to connect the observed movement to the proposed control rule and state the conditions under which that rule was tested.
The 2015 phantom-road experiment by Ware and colleagues broadcast traffic noise into a roadless landscape. This separated noise from several other features of a real road. The authors reported avoidance by 31% of the bird community and reduced body condition among birds that remained, with species differences in stopover responses. Complementary laboratory work implicated a feeding–vigilance mechanism.
The result illustrates why presence alone is insufficient evidence of an unaffected habitat. An animal can remain while experiencing a cost. It also illustrates the value of manipulating one proposed cause rather than treating proximity to a complex disturbance as proof of its mechanism.
An original follow-up planning question is: what should a habitat manager measure after changing a sound source? Counts alone could miss altered behaviour, while behaviour alone might not establish a lasting population effect. A useful monitoring plan links exposure, individual response and the outcome the intervention aims to improve.
Write those outcomes before collecting data. For a hypothetical feeding study, they might include time available for feeding, successful feeding events per observation period and a separately justified condition measure. Keep observation effort consistent. If observers spend longer watching the disturbed area, raw event totals are not comparable.
Do not extend one study’s percentage to all birds or all roads. Species, season, habitat and exposure pattern can change the result. The transferable insight is the experimental structure: isolate a sensory disturbance, measure its consequences and retain the boundary between the tested setting and a wider prediction.
Suppose an animal does not approach a stimulus. It might not detect it. It might detect it but not distinguish it from the background. It might recognise it and choose not to approach. It might be occupied with another task. A single absence of movement cannot decide among these explanations.
Use a sequence of questions. First, establish that the apparatus delivers the intended physical stimulus at the receiver’s position. Second, test an appropriate response under a condition in which the animal can perform the task. Third, alter the feature of interest while controlling plausible alternatives. Finally, examine whether the effect persists under ecologically relevant conditions.
Randomise treatment order where possible. Otherwise fatigue, learning or changing motivation can look like an effect of stimulus strength. Give repeated measurements their correct role: twenty trials from one animal are not twenty independent animals. Report the number of individuals, the number of trials per individual and the structure used in analysis.
Keep scoring criteria operational. “Looked confused” is a weak outcome unless translated into observable actions. A predefined turn angle, response latency or path deviation can be more reproducible. Blind scoring to the treatment when feasible, and preserve the original observations so that another reviewer can check borderline cases.
Avoid a false choice between naturalism and control. A controlled study can isolate a mechanism; a field study can test whether that mechanism matters in a complex setting. Confidence grows when the evidence connects these levels. A beautiful field observation cannot replace a missing control, and a clean laboratory effect cannot alone establish population consequences.
Consider an invented study of a visual food-location task. The same ten animals complete matched sessions before and after a background change. In the first condition, 80 of 100 opportunities produce a correct choice. In the second, 60 of 100 do. The observed success rates are 80% and 60%, a decrease of 20 percentage points or 25% relative to the initial rate.
These totals describe the observations but do not settle the cause. If every animal always experienced the first condition before the second, declining motivation could explain part of the difference. If a few animals produced most opportunities, the pooled rate may hide individual variation. If the background change also affected temperature, a visual interpretation needs further controls.
Now add the contrast calculation from chapter 4. Suppose the target remained at 12 units while the background changed from 10 to 11. The measured contrast fell from 0.20 to about 0.091. That is consistent with a contrast-based hypothesis, but consistency is not proof. A stronger design would vary contrast independently of plausible confounders and test the prediction across individuals.
Transfer question
In a new setting, target and background both double from 12 and 10 to 24 and 20 units. A colleague predicts the same decline in performance because the environment became brighter. Is that prediction supported by the contrast calculation?
Answer and reasoning
No. The original contrast was 0.20 and the new contrast is also 0.20. The calculation does not predict the same contrast-driven decline. It also does not guarantee unchanged performance: absolute light availability, adaptation or other features could matter. The correct conclusion preserves what the model says and identifies what it leaves unresolved.
This is the transfer habit developed in Education: learning that transfers into life. Change an assumption deliberately, recalculate the relevant relationship and resist carrying over an answer merely because the new story sounds similar.
“The species is present, so the disturbance is harmless.” Repair this by measuring the relevant performance or condition outcome. Presence answers where an organism was detected; it does not establish that the setting has no costs. Conversely, absence from a short survey does not establish that the species cannot use the habitat.
“The sensor detected it, so the animal did.” Repair this by checking the receiver’s sensitivity, the stimulus at its position and behavioural evidence. Instrument settings can create apparent clarity that an animal does not possess. Report calibration and limitations rather than substituting a colourful display for a biological measurement.
“The brighter treatment should always work better.” Repair this by separating absolute intensity, contrast, spectrum and adaptation. The worked contrast example shows why a brighter target can retain the same relative contrast. A field intervention needs a clearly stated mechanism, not a single adjective.
“A response proves the stimulus has an adaptive function.” Repair this by distinguishing immediate response from evolutionary explanation. A response may be a side effect, a learned association or a consequence of a rule that is unreliable in the current setting. Establishing fitness consequences and evolutionary history requires additional evidence.
“No significant difference means the treatments are identical.” Repair this by considering uncertainty, sample size and the range of effects compatible with the data. A study may be unable to resolve a biologically important difference. Report what precision was achieved and what decision the evidence can support.
“One species represents all animals.” Repair this by identifying the feature needed for transfer: receptor properties, habitat, task, life stage or exposure. Similar-looking organs do not guarantee identical performance. A useful comparison states which relationship is shared and which parameters need new measurements.
Begin with observation that does not disturb wildlife. Describe a scene’s light sources, backgrounds, barriers and changing sounds. Record the time and conditions. Separate your own sensory experience from claims about another animal’s perception. A quiet scene for a person is not automatically quiet in every biologically relevant frequency band.
Next, use a paper model. Work through the contrast and acoustic-range examples with altered numbers. Predict the direction of change before calculating. Check units and denominators. If your prediction fails, inspect the assumptions rather than adding a story after the arithmetic.
Then compare a research abstract with its actual claim. Identify the species or community, the manipulated variable, the measured response and the stated limitation. Ask whether an article about the study has extended the conclusion beyond that boundary. This exercise turns reading into a test of explanation rather than a collection of striking facts.
Students interested in further study can connect biology with optics, acoustics, statistics and experimental design. Practical work may involve behavioural observation, calibrated recording, image analysis or environmental measurement. Different roles require different training; familiarity with a recording tool alone does not qualify someone to interpret every biological result.
For real-world planning, begin with a specific objective such as preserving a dark movement corridor or reducing interference with a defined acoustic task. Identify the relevant organisms and exposure conditions, compare feasible interventions and monitor the outcome. The examples here support reasoning; they do not supply universal safe lighting levels or species-independent noise limits.
Is sensory ecology just animal communication?
No. Communication is one important part. Sensory ecology also examines information from the physical environment and incidental cues from other organisms. A cue used to maintain orientation need not be a message produced by a sender.
Does a more sensitive sense always improve survival?
Not necessarily. Sensitivity is one performance dimension. Discrimination, speed, energy costs and the consequences of false responses also matter. A claim about survival needs ecological evidence rather than a ranking of receptor sensitivity alone.
Can a photograph show what an animal sees?
A photograph can document a scene, but ordinary camera output is not a direct reconstruction of another species’ visual experience. Illumination, spectral sensitivity, exposure, processing and display all matter. A justified model needs relevant measurements and explicit assumptions.
Is sensory pollution only noise and light?
Those are prominent examples, but the broader question concerns alterations to usable sensory information. Assess the modality and mechanism in the particular case. Do not label every environmental change harmful without testing exposure and consequence.
Why keep using fictional numbers?
They make the arithmetic reproducible without presenting an invented species record as fact. The contrast, detection-range and response-rate examples isolate relationships. Real applications need measured parameters, uncertainty estimates and a design suited to the biological question.
What is the quickest test of an explanation?
Ask what the receiver actually encountered, what changed in its response and which alternative explanation was tested. Then ask what evidence connects that response to an ecological outcome. A missing link is a useful next question, not permission to assume the answer.
Use Insect Communication for the distinctions among chemical, acoustic, visual and tactile signals. Use Bioacoustics for recording and acoustic inference. These routes deepen particular modalities while retaining the need for receiver evidence.
The specialist article Bioluminescence, Deep-Sea Light and Adaptation Evidence follows light-producing chemistry, propagation and ecological explanations. Read it alongside the visual contrast chapter here: producing light, transmitting light and extracting useful information from light are connected but different problems.
Return to the Gold Standard master to explore neighbouring subjects, or use How X Works to follow a mechanism into another field. The most useful next article depends on the missing link in your explanation: physical transmission, receiver biology, behavioural response or ecological consequence.
The research cases in this guide are deliberately bounded. Fabian and colleagues’ 2024 insect-flight paper supplies a tested orientation mechanism near lights. Ware and colleagues’ 2015 phantom-road paper supplies an experimental route from noise exposure toward habitat consequences. They are examples of evidence, not universal numerical rules for all species.
For your own final check, choose one environmental change and write a prediction that could fail. Name the receiver, the physical variable, the predicted response and a meaningful comparison. Include one alternative explanation and one result that would make you revise your account. That is how sensory ecology turns an appealing description of another animal’s world into a testable explanation.